🤖 ModelBytes Digest
Sunday, August 16, 2026
━━━ OPEN FRONTIER 🔓
Qwen3.8-27B-XYZ
ML model...
Released: 2026-08-14 | License: apache-2.0 | Params: 27B
📦 Open weights · HF
→ Source
Total: 1 items tracked today
Sunday, August 16, 2026
━━━ OPEN FRONTIER 🔓
Qwen3.8-27B-XYZ
ML model...
Released: 2026-08-14 | License: apache-2.0 | Params: 27B
📦 Open weights · HF
→ Source
Total: 1 items tracked today
🤖 ModelBytes Digest
Monday, August 17, 2026
Today's batch shows a surge of open‑weight, domain‑specific models aimed at safety, security and visual tasks.
━━━ OPEN FRONTIER 🔓
Qwen 3.8 27B — 27B uncensored LLM released with full weight access, targeting aggressive generation. 27B params. 📦 Open weights. → Source
━━━ SPECIALIZED 🎯
Shieldstral — 3B open‑weights multimodal safety classifier that outperforms models up to 7× its size. 3B params. 📦 Open weights. → Source
Antares — Highly efficient open‑weight models for vulnerability localization, cutting compute needs dramatically. 📦 Open weights. → Source
Muse Glimmer — 30B image‑text‑to‑text model optimized for single‑GPU agentic tasks on consumer hardware. 30B params. Apache‑2.0 license. 📦 Open weights · HF. → Source
Wan2.2 Animate 2 14B Diffusers — 14B diffusion model for high‑quality animation generation. 14B params. Apache‑2.0 license. 📦 Open weights · HF. → Source
EVIE Preview 4.5B — 4.5B visual‑document‑retrieval model for fast multimodal search. 4.5B params. Apache‑2.0 license. 📦 Open weights · HF. → Source
📊 Surfaced 6 · scanned 3 today
Monday, August 17, 2026
Today's batch shows a surge of open‑weight, domain‑specific models aimed at safety, security and visual tasks.
━━━ OPEN FRONTIER 🔓
Qwen 3.8 27B — 27B uncensored LLM released with full weight access, targeting aggressive generation. 27B params. 📦 Open weights. → Source
━━━ SPECIALIZED 🎯
Shieldstral — 3B open‑weights multimodal safety classifier that outperforms models up to 7× its size. 3B params. 📦 Open weights. → Source
Antares — Highly efficient open‑weight models for vulnerability localization, cutting compute needs dramatically. 📦 Open weights. → Source
Muse Glimmer — 30B image‑text‑to‑text model optimized for single‑GPU agentic tasks on consumer hardware. 30B params. Apache‑2.0 license. 📦 Open weights · HF. → Source
Wan2.2 Animate 2 14B Diffusers — 14B diffusion model for high‑quality animation generation. 14B params. Apache‑2.0 license. 📦 Open weights · HF. → Source
EVIE Preview 4.5B — 4.5B visual‑document‑retrieval model for fast multimodal search. 4.5B params. Apache‑2.0 license. 📦 Open weights · HF. → Source
📊 Surfaced 6 · scanned 3 today
🤖 ModelBytes Digest
Tuesday, August 18, 2026
Open-weight releases are compressing frontier capability into smaller, task-specific packages this week — builders get coding, multimodal safety, and UI understanding without the closed API tax.
━━━ OPEN FRONTIER 🔓
Muse Glimmer — Meta's smaller open-weight model, positioned as a lightweight alternative to larger rivals. Released Aug 10. 📦 Open weights. → Source
Qwen 3.8 27B — Latest Qwen release. 27B params. Released Aug 14. 📦 Open weights. → Source
━━━ CLOSED FRONTIER 🔒
Gemini 3.7 Flash — Google's newest Flash-tier model, iterating on speed and cost. Released Aug 13. ⚡ API live. → Source
━━━ SPECIALIZED 🎯
Shieldstral 1.0 3B — Open-weights policy-adaptive multimodal safety classifier matching models 7× its size. 3B params, open weights. Released Aug 7. 📦 Open weights. → Source
UI-Mate 27B — Image-text-to-text model for UI understanding. 27B params, Apache-2.0. Released Aug 14. 📦 Open weights · HF. → Source
UI-Mate 9B — Smaller image-text-to-text model for UI understanding. 9B params, Apache-2.0. Released Aug 14. 📦 Open weights · HF. → Source
📊 Surfaced 6 · scanned 2 today
Tuesday, August 18, 2026
Open-weight releases are compressing frontier capability into smaller, task-specific packages this week — builders get coding, multimodal safety, and UI understanding without the closed API tax.
━━━ OPEN FRONTIER 🔓
Muse Glimmer — Meta's smaller open-weight model, positioned as a lightweight alternative to larger rivals. Released Aug 10. 📦 Open weights. → Source
Qwen 3.8 27B — Latest Qwen release. 27B params. Released Aug 14. 📦 Open weights. → Source
━━━ CLOSED FRONTIER 🔒
Gemini 3.7 Flash — Google's newest Flash-tier model, iterating on speed and cost. Released Aug 13. ⚡ API live. → Source
━━━ SPECIALIZED 🎯
Shieldstral 1.0 3B — Open-weights policy-adaptive multimodal safety classifier matching models 7× its size. 3B params, open weights. Released Aug 7. 📦 Open weights. → Source
UI-Mate 27B — Image-text-to-text model for UI understanding. 27B params, Apache-2.0. Released Aug 14. 📦 Open weights · HF. → Source
UI-Mate 9B — Smaller image-text-to-text model for UI understanding. 9B params, Apache-2.0. Released Aug 14. 📦 Open weights · HF. → Source
📊 Surfaced 6 · scanned 2 today
🤖 ModelBytes Digest
Wednesday, August 19, 2026
Open-weight releases are splitting into compact agentic models and frontier-scale reasoning, so builders can now choose cost/performance without leaving the open ecosystem.
━━━ OPEN FRONTIER 🔓
Muse Spark 1.2 — Meta's latest open-weight model, open weights per Zuckerberg. 📦 Open weights. → Source
Qwen3.8 Max — Largest open-weight release in history, per local-ai-zone. 📦 Open weights. → Source
Nemotron 3.5 Lightning — Compact open agentic model in the same wave as Muse Glimmer and Qwen3.8-27B. 📦 Open weights. → Source
━━━ CLOSED FRONTIER 🔒
GLM 5.3 — Large-scale reasoning model for complex software engineering and long-horizon agent tasks. Context 1.0M · License Closed/API · Pricing $1.40/$4.40 per 1M · Released Aug 18. ⚡ API live · OpenRouter. → Source
📊 Surfaced 4 · scanned 3 today
Wednesday, August 19, 2026
Open-weight releases are splitting into compact agentic models and frontier-scale reasoning, so builders can now choose cost/performance without leaving the open ecosystem.
━━━ OPEN FRONTIER 🔓
Muse Spark 1.2 — Meta's latest open-weight model, open weights per Zuckerberg. 📦 Open weights. → Source
Qwen3.8 Max — Largest open-weight release in history, per local-ai-zone. 📦 Open weights. → Source
Nemotron 3.5 Lightning — Compact open agentic model in the same wave as Muse Glimmer and Qwen3.8-27B. 📦 Open weights. → Source
━━━ CLOSED FRONTIER 🔒
GLM 5.3 — Large-scale reasoning model for complex software engineering and long-horizon agent tasks. Context 1.0M · License Closed/API · Pricing $1.40/$4.40 per 1M · Released Aug 18. ⚡ API live · OpenRouter. → Source
📊 Surfaced 4 · scanned 3 today
🤖 ModelBytes Digest
Thursday, August 20, 2026
Today's releases are narrow: Tencent ships two translation-specific APIs while Z.Ai iterates GLM with a Turbo variant, so builders get cheaper specialized calls rather than a new general frontier.
━━━ CLOSED FRONTIER 🔒
GLM-5.2 Turbo — Z.AI's faster GLM-5.2 variant, released Aug 17. Released Aug 17. → Source
━━━ SPECIALIZED 🎯
Hy-MT2 1.8B — Compact translation model covering 33 language pairs plus five Chinese dialect and minority-language pairs. Released Aug 20 · 8k context · Closed/API · $0.04/$0.18 per 1M tokens. ⚡ API live · OpenRouter. → Source
Hy-MT2 30B A3B — Tencent's flagship translation model with the same 33+5 language coverage and structured workflows. Released Aug 20 · 8k context · Closed/API · $0.07/$0.29 per 1M tokens. ⚡ API live · OpenRouter. → Source
📊 Surfaced 3 · scanned 3 today
Thursday, August 20, 2026
Today's releases are narrow: Tencent ships two translation-specific APIs while Z.Ai iterates GLM with a Turbo variant, so builders get cheaper specialized calls rather than a new general frontier.
━━━ CLOSED FRONTIER 🔒
GLM-5.2 Turbo — Z.AI's faster GLM-5.2 variant, released Aug 17. Released Aug 17. → Source
━━━ SPECIALIZED 🎯
Hy-MT2 1.8B — Compact translation model covering 33 language pairs plus five Chinese dialect and minority-language pairs. Released Aug 20 · 8k context · Closed/API · $0.04/$0.18 per 1M tokens. ⚡ API live · OpenRouter. → Source
Hy-MT2 30B A3B — Tencent's flagship translation model with the same 33+5 language coverage and structured workflows. Released Aug 20 · 8k context · Closed/API · $0.07/$0.29 per 1M tokens. ⚡ API live · OpenRouter. → Source
📊 Surfaced 3 · scanned 3 today
🤖 ModelBytes Digest
Friday, August 21, 2026
Open-weight multimodal and coding models are closing the gap with closed frontier while a 3B safety classifier makes moderation cheap — builders can now assemble vision, coding, and safety from open components without API lock-in.
━━━ OPEN FRONTIER 🔓
GLM-5.3 — Post-training-only upgrade on the 743B GLM-5.2 base claims a 50% coding jump and top open Terminal-Bench scores. 743B total params · open weights · Released Aug 14. 📦 Open weights. → Source
DeepSeek V4 Flash Vision Exp — Experimental vision-enabled V4 Flash matches text performance and closes most of the gap to Opus-4.8 on multimodal agent tasks. 284B total / 13B active · 1M context · MIT · $0.22/$0.66 per 1M · Released Aug 21. ⚡ API live · OpenRouter. → Source
Ling 3.0 Flash Base — 127.5B open-weights text-generation base model under MIT, a clean pretrained foundation for further tuning. 127.5B total params · MIT · Released Aug 11. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Ox Alpha — Reasoning model built for coding, sustained agentic work, and production workloads with a 1M-token context. 1M context · Closed/API · Free · Released Aug 20. ⚡ API live · OpenRouter. → Source
📊 Surfaced 4 · scanned 10 today
Friday, August 21, 2026
Open-weight multimodal and coding models are closing the gap with closed frontier while a 3B safety classifier makes moderation cheap — builders can now assemble vision, coding, and safety from open components without API lock-in.
━━━ OPEN FRONTIER 🔓
GLM-5.3 — Post-training-only upgrade on the 743B GLM-5.2 base claims a 50% coding jump and top open Terminal-Bench scores. 743B total params · open weights · Released Aug 14. 📦 Open weights. → Source
DeepSeek V4 Flash Vision Exp — Experimental vision-enabled V4 Flash matches text performance and closes most of the gap to Opus-4.8 on multimodal agent tasks. 284B total / 13B active · 1M context · MIT · $0.22/$0.66 per 1M · Released Aug 21. ⚡ API live · OpenRouter. → Source
Ling 3.0 Flash Base — 127.5B open-weights text-generation base model under MIT, a clean pretrained foundation for further tuning. 127.5B total params · MIT · Released Aug 11. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Ox Alpha — Reasoning model built for coding, sustained agentic work, and production workloads with a 1M-token context. 1M context · Closed/API · Free · Released Aug 20. ⚡ API live · OpenRouter. → Source
📊 Surfaced 4 · scanned 10 today
🤖 ModelBytes Digest
Saturday, August 22, 2026
━━━ OPEN FRONTIER 🔓
Ornith 1.5 35B A3B — MIT-licensed 35B-parameter text model; the A3B suffix suggests a sparse architecture but active params are unconfirmed. Released Aug 22 · 35B total params · MIT license. 📦 Open weights · HF. → Source
📊 Surfaced 1 · scanned 2 today
Saturday, August 22, 2026
━━━ OPEN FRONTIER 🔓
Ornith 1.5 35B A3B — MIT-licensed 35B-parameter text model; the A3B suffix suggests a sparse architecture but active params are unconfirmed. Released Aug 22 · 35B total params · MIT license. 📦 Open weights · HF. → Source
📊 Surfaced 1 · scanned 2 today
🤖 ModelBytes Digest
Sunday, August 23, 2026
━━━ SPECIALIZED 🎯
Hy-MT2 7B — Tencent's 7B translation model covers 33 language pairs plus five Chinese dialect and minority-language pairs, with structured delimiter-based workflows for clean extraction. Released Aug 19 · 7B params · 8k context · Closed/API · $0.07/$0.29 per 1M. ⚡ API live · OpenRouter. → Source
📊 Surfaced 1 · scanned 1 today
Sunday, August 23, 2026
━━━ SPECIALIZED 🎯
Hy-MT2 7B — Tencent's 7B translation model covers 33 language pairs plus five Chinese dialect and minority-language pairs, with structured delimiter-based workflows for clean extraction. Released Aug 19 · 7B params · 8k context · Closed/API · $0.07/$0.29 per 1M. ⚡ API live · OpenRouter. → Source
📊 Surfaced 1 · scanned 1 today
🤖 ModelBytes Digest
Tuesday, August 25, 2026
━━━ CLOSED FRONTIER 🔒
Grok 4.6 — xAI's most capable model, built for coding, agentic tasks, and knowledge work. Released Aug 21. ⚡ API live. → Source
📊 Surfaced 1 · scanned 10 today
Tuesday, August 25, 2026
━━━ CLOSED FRONTIER 🔒
Grok 4.6 — xAI's most capable model, built for coding, agentic tasks, and knowledge work. Released Aug 21. ⚡ API live. → Source
📊 Surfaced 1 · scanned 10 today
🤖 ModelBytes Digest
Wednesday, August 26, 2026
Builders get a two-track week: open-weight models pushing into frontier-scale parameter counts, and specialized tools that skip the API entirely.
━━━ OPEN FRONTIER 🔓
Qwen3.8 Flash Next — Image-text-to-text at 180B total params, aimed at multimodal tasks without a separate vision tower. 180B params · Released Aug 26 · 📦 Open weights · HF → Source
GLM 5.3 Flash — MIT-licensed 321B text-generation model, the open-weight counterpart to the closed GLM 5.3 line. 321.3B params · MIT · Released Aug 26 · 📦 Open weights · HF → Source
━━━ SPECIALIZED 🎯
MiniMax Music3 — Open-weights music model that generates complete five-minute songs from lyrics plus a structured caption, not just short clips. Released Aug 17 · 📦 Open weights → Source
📊 Surfaced 3 · scanned 5 today
Wednesday, August 26, 2026
Builders get a two-track week: open-weight models pushing into frontier-scale parameter counts, and specialized tools that skip the API entirely.
━━━ OPEN FRONTIER 🔓
Qwen3.8 Flash Next — Image-text-to-text at 180B total params, aimed at multimodal tasks without a separate vision tower. 180B params · Released Aug 26 · 📦 Open weights · HF → Source
GLM 5.3 Flash — MIT-licensed 321B text-generation model, the open-weight counterpart to the closed GLM 5.3 line. 321.3B params · MIT · Released Aug 26 · 📦 Open weights · HF → Source
━━━ SPECIALIZED 🎯
MiniMax Music3 — Open-weights music model that generates complete five-minute songs from lyrics plus a structured caption, not just short clips. Released Aug 17 · 📦 Open weights → Source
📊 Surfaced 3 · scanned 5 today
🤖 ModelBytes Digest
Thursday, August 27, 2026
Today's releases push multimodal and speech-to-text into default building blocks: a 1M-context Qwen flash for agentic UI work, an open 9B image-text model, and GA transcription from Gemini.
━━━ OPEN FRONTIER 🔓
Qwen3.8 Flash — Multimodal reasoning for coding, agentic workflows, visual understanding, and document/codebase analysis. Released Aug 26. Context 1.0M; $0.15/$0.47 per 1M. ⚡ API live · OpenRouter. → Source
━━━ SPECIALIZED 🎯
UI-Venus 2 9B — Open 9B image-text-to-text model for UI-centric multimodal workflows. Released Aug 26. 9B params; Apache-2.0. 📦 Open weights · HF. → Source
Gemini 3.5 Transcribe — Two dedicated speech-to-text models based on Gemini audio understanding, now generally available. Released Aug 26. GA. ⚡ API live · Google AI. → Source
📊 Surfaced 3 · scanned 8 today
Thursday, August 27, 2026
Today's releases push multimodal and speech-to-text into default building blocks: a 1M-context Qwen flash for agentic UI work, an open 9B image-text model, and GA transcription from Gemini.
━━━ OPEN FRONTIER 🔓
Qwen3.8 Flash — Multimodal reasoning for coding, agentic workflows, visual understanding, and document/codebase analysis. Released Aug 26. Context 1.0M; $0.15/$0.47 per 1M. ⚡ API live · OpenRouter. → Source
━━━ SPECIALIZED 🎯
UI-Venus 2 9B — Open 9B image-text-to-text model for UI-centric multimodal workflows. Released Aug 26. 9B params; Apache-2.0. 📦 Open weights · HF. → Source
Gemini 3.5 Transcribe — Two dedicated speech-to-text models based on Gemini audio understanding, now generally available. Released Aug 26. GA. ⚡ API live · Google AI. → Source
📊 Surfaced 3 · scanned 8 today
🤖 ModelBytes Digest
Friday, August 28, 2026
Today's releases split between massive open-weight generalists (Hy4 Preview at 780B, Granite 4.2 local reasoning) and narrow specialized tools (finance, diarization, time-series), so builders can pair a local reasoning core with cheap domain-specific APIs.
━━━ OPEN FRONTIER 🔓
Hy4 Preview — Tencent's 780B open-weight text-generation model under Apache 2.0, a large generalist for custom deployment. 780B params · Apache 2.0 · Released Aug 27. 📦 Open weights · HF. → Source
━━━ SPECIALIZED 🎯
Ling 3.0 Flash Fin — Finance-focused MoE with 5.1B active params out of 124B, 262k context, free API for investment workflows. 5.1B active / 124B total · 262k context · Closed/API · FREE · Released Aug 27. ⚡ API live · OpenRouter. → Source
Nemotron 3 Diarization Preview — Voice-activity-detection model for speaker diarization, a building block for audio pipelines. Released Aug 24. 📦 Open weights · HF. → Source
TimesFM 3.0 — Time-series forecasting model at 331M params, small enough to embed in analytics. 331M params · Released Aug 24. 📦 Open weights · HF. → Source
━━━ LOCAL 🏠
Granite 4.2 — Open-weight reasoning family in 3B, 8B, and 30B sizes for local and enterprise deployment under Apache 2.0. 3B/8B/30B · Apache 2.0 · Released Aug 25. 📦 Open weights. → Source
📊 Surfaced 5 · scanned 8 today
Friday, August 28, 2026
Today's releases split between massive open-weight generalists (Hy4 Preview at 780B, Granite 4.2 local reasoning) and narrow specialized tools (finance, diarization, time-series), so builders can pair a local reasoning core with cheap domain-specific APIs.
━━━ OPEN FRONTIER 🔓
Hy4 Preview — Tencent's 780B open-weight text-generation model under Apache 2.0, a large generalist for custom deployment. 780B params · Apache 2.0 · Released Aug 27. 📦 Open weights · HF. → Source
━━━ SPECIALIZED 🎯
Ling 3.0 Flash Fin — Finance-focused MoE with 5.1B active params out of 124B, 262k context, free API for investment workflows. 5.1B active / 124B total · 262k context · Closed/API · FREE · Released Aug 27. ⚡ API live · OpenRouter. → Source
Nemotron 3 Diarization Preview — Voice-activity-detection model for speaker diarization, a building block for audio pipelines. Released Aug 24. 📦 Open weights · HF. → Source
TimesFM 3.0 — Time-series forecasting model at 331M params, small enough to embed in analytics. 331M params · Released Aug 24. 📦 Open weights · HF. → Source
━━━ LOCAL 🏠
Granite 4.2 — Open-weight reasoning family in 3B, 8B, and 30B sizes for local and enterprise deployment under Apache 2.0. 3B/8B/30B · Apache 2.0 · Released Aug 25. 📦 Open weights. → Source
📊 Surfaced 5 · scanned 8 today
🤖 ModelBytes Digest
Saturday, August 29, 2026
━━━ ALSO TRACKED
• glm-5.3-flash:batch (openrouter)
Total: 1 items tracked today
Saturday, August 29, 2026
━━━ ALSO TRACKED
• glm-5.3-flash:batch (openrouter)
Total: 1 items tracked today
🤖 ModelBytes Digest
Sunday, August 30, 2026
Today’s only fresh release is a 30B open-weight model aimed at 20GB VRAM, continuing the local-first push.
━━━ LOCAL 🏠
Meta Muse Glimmer — 30B open-weight model sized for consumer GPUs, fitting in 20GB VRAM for local deployment. 30B params, open weights. 📦 Open weights. → Source
📊 Surfaced 1 · scanned 2 today
Sunday, August 30, 2026
Today’s only fresh release is a 30B open-weight model aimed at 20GB VRAM, continuing the local-first push.
━━━ LOCAL 🏠
Meta Muse Glimmer — 30B open-weight model sized for consumer GPUs, fitting in 20GB VRAM for local deployment. 30B params, open weights. 📦 Open weights. → Source
📊 Surfaced 1 · scanned 2 today
🤖 ModelBytes Digest
Monday, August 31, 2026
Today’s split: closed frontier polish in Grok 4.1 versus open-weight specialization for Indian scripts and long-context handling — builders get cheaper local options without waiting for the next giant.
━━━ OPEN FRONTIER 🔓
Gnani Evon — Open-weight model with a tokenizer rebuilt for Indian scripts, targeting lower AI costs for enterprise agents. Open weights. Released Aug 29. 📦 Open weights. → Source
ContextPilot 8B — Text-generation model at 8B total params, part of a family for context handling. 8B params. Released Aug 27. 📦 Open weights · HF. → Source
ContextPilot 14B — Larger 14B sibling for context-heavy tasks. 14B params. Released Aug 27. 📦 Open weights · HF. → Source
ContextPilot E4B — 7.9B efficient variant, likely for lower-resource deployment. 7.9B params. Released Aug 27. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Grok 4.1 — Lower hallucination rate and API access at competitive pricing, positioned against Gemini 3. ⚡ API live. → Source
📊 Surfaced 5 · scanned 4 today
Monday, August 31, 2026
Today’s split: closed frontier polish in Grok 4.1 versus open-weight specialization for Indian scripts and long-context handling — builders get cheaper local options without waiting for the next giant.
━━━ OPEN FRONTIER 🔓
Gnani Evon — Open-weight model with a tokenizer rebuilt for Indian scripts, targeting lower AI costs for enterprise agents. Open weights. Released Aug 29. 📦 Open weights. → Source
ContextPilot 8B — Text-generation model at 8B total params, part of a family for context handling. 8B params. Released Aug 27. 📦 Open weights · HF. → Source
ContextPilot 14B — Larger 14B sibling for context-heavy tasks. 14B params. Released Aug 27. 📦 Open weights · HF. → Source
ContextPilot E4B — 7.9B efficient variant, likely for lower-resource deployment. 7.9B params. Released Aug 27. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Grok 4.1 — Lower hallucination rate and API access at competitive pricing, positioned against Gemini 3. ⚡ API live. → Source
📊 Surfaced 5 · scanned 4 today
🤖 ModelBytes Digest
Tuesday, September 01, 2026
Today's releases split between a faster closed reasoning API and sparse open-weight additions, so builders get a cheap diffusion-LLM option but little new local horsepower.
━━━ OPEN FRONTIER 🔓
InternLumina U2 — New Apache-2.0 open-weight release from InternLM with no published parameter count or task focus yet. License Apache-2.0 · Released Sep 1. 📦 Open weights · HF. → Source
Granite 4.2 8B — IBM's dense reasoning model gets an 8B API variant, adding a cheaper option for math, code generation, and agentic workflows. Context 131k · $0.10/$0.15 per 1M · Released Aug 31. ⚡ API live · OpenRouter. → Source
━━━ CLOSED FRONTIER 🔒
Mercury 2.5 Preview — Diffusion LLM that produces and refines multiple tokens in parallel, targeting the fastest reasoning latency. Closed/API · Context 260k · $0.04/$0.15 per 1M · Released Aug 31. ⚡ API live · OpenRouter. → Source
📊 Surfaced 3 · scanned 11 today
Tuesday, September 01, 2026
Today's releases split between a faster closed reasoning API and sparse open-weight additions, so builders get a cheap diffusion-LLM option but little new local horsepower.
━━━ OPEN FRONTIER 🔓
InternLumina U2 — New Apache-2.0 open-weight release from InternLM with no published parameter count or task focus yet. License Apache-2.0 · Released Sep 1. 📦 Open weights · HF. → Source
Granite 4.2 8B — IBM's dense reasoning model gets an 8B API variant, adding a cheaper option for math, code generation, and agentic workflows. Context 131k · $0.10/$0.15 per 1M · Released Aug 31. ⚡ API live · OpenRouter. → Source
━━━ CLOSED FRONTIER 🔒
Mercury 2.5 Preview — Diffusion LLM that produces and refines multiple tokens in parallel, targeting the fastest reasoning latency. Closed/API · Context 260k · $0.04/$0.15 per 1M · Released Aug 31. ⚡ API live · OpenRouter. → Source
📊 Surfaced 3 · scanned 11 today
🤖 ModelBytes Digest
Wednesday, September 02, 2026
Today's pattern is incremental frontier upgrades for agentic coding and long-horizon work, plus one small open-weight model for local tinkering.
━━━ CLOSED FRONTIER 🔒
Claude Fable 5.1 — Successor to Fable 5 with the biggest gains in agentic coding, long-running workflows, and knowledge work; same price but cheaper cache reads. 1M context, closed API, $10/$50 per 1M. Released Sep 1. ⚡ API live · OpenRouter. → Source
Claude Mythos 5.1 — Anthropic's companion reasoning-vision model launched alongside Fable 5.1. Closed API. Released Sep 1. ⚡ API live · Anthropic. → Source
Gemini 3.8 Flash — Google's most intelligent Flash model, with significant gains over 3.7 Flash in software engineering, agentic tasks, and multi-step reasoning. 1M context, closed API, $0.75/$3.75 per 1M. Released Sep 2. ⚡ API live · OpenRouter. → Source
━━━ LOCAL 🏠
tiny-aya-l2-thinker — Cohere's 3.4B open-weight text-generation model under a non-commercial license, small enough for local experimentation. 3.4B total params, CC BY-NC 4.0. Released Sep 2. 📦 Open weights · HF. → Source
📊 Surfaced 4 · scanned 10 today
Wednesday, September 02, 2026
Today's pattern is incremental frontier upgrades for agentic coding and long-horizon work, plus one small open-weight model for local tinkering.
━━━ CLOSED FRONTIER 🔒
Claude Fable 5.1 — Successor to Fable 5 with the biggest gains in agentic coding, long-running workflows, and knowledge work; same price but cheaper cache reads. 1M context, closed API, $10/$50 per 1M. Released Sep 1. ⚡ API live · OpenRouter. → Source
Claude Mythos 5.1 — Anthropic's companion reasoning-vision model launched alongside Fable 5.1. Closed API. Released Sep 1. ⚡ API live · Anthropic. → Source
Gemini 3.8 Flash — Google's most intelligent Flash model, with significant gains over 3.7 Flash in software engineering, agentic tasks, and multi-step reasoning. 1M context, closed API, $0.75/$3.75 per 1M. Released Sep 2. ⚡ API live · OpenRouter. → Source
━━━ LOCAL 🏠
tiny-aya-l2-thinker — Cohere's 3.4B open-weight text-generation model under a non-commercial license, small enough for local experimentation. 3.4B total params, CC BY-NC 4.0. Released Sep 2. 📦 Open weights · HF. → Source
📊 Surfaced 4 · scanned 10 today
🤖 ModelBytes Digest
Thursday, September 03, 2026
Today's pattern is looser access: Meta teases open weights for Muse Spark while shipping a closed API tier, Anthropic cuts Fable/Mythos prices, and open labs push small permissive multimodal and reasoning models.
━━━ OPEN FRONTIER 🔓
Qwen Drive 1.0 4B — Small open multimodal model for image+text tasks where a 4B footprint is enough. 4B params, Apache-2.0. Released Aug 27. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Muse Spark 1.3 — Meta's multimodal reasoning model for long-running agentic and coding workflows; open weights teased, with a cheaper contributor tier at $0.10/$0.20 per 1M. 1M context, closed API, $1.25/$4.25 per 1M. Released Sep 2. ⚡ API live · OpenRouter. → Source
Gemini 3.8 Flash Cyber — Cyber variant of the 3.8 Flash line, announced alongside the base model. Released Sep 2. → Source
━━━ SPECIALIZED 🎯
MobileVLA R1 — Mobile vision-language-action model from BAAI, CC-BY-SA-4.0. Released Sep 3. 📦 Open weights · HF. → Source
Recon2Reason Reasoning 4B — 4B image-text-to-text reasoning model, Apache-2.0. 4B params, Apache-2.0. Released Sep 3. 📦 Open weights · HF. → Source
📊 Surfaced 5 · scanned 9 today
Thursday, September 03, 2026
Today's pattern is looser access: Meta teases open weights for Muse Spark while shipping a closed API tier, Anthropic cuts Fable/Mythos prices, and open labs push small permissive multimodal and reasoning models.
━━━ OPEN FRONTIER 🔓
Qwen Drive 1.0 4B — Small open multimodal model for image+text tasks where a 4B footprint is enough. 4B params, Apache-2.0. Released Aug 27. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Muse Spark 1.3 — Meta's multimodal reasoning model for long-running agentic and coding workflows; open weights teased, with a cheaper contributor tier at $0.10/$0.20 per 1M. 1M context, closed API, $1.25/$4.25 per 1M. Released Sep 2. ⚡ API live · OpenRouter. → Source
Gemini 3.8 Flash Cyber — Cyber variant of the 3.8 Flash line, announced alongside the base model. Released Sep 2. → Source
━━━ SPECIALIZED 🎯
MobileVLA R1 — Mobile vision-language-action model from BAAI, CC-BY-SA-4.0. Released Sep 3. 📦 Open weights · HF. → Source
Recon2Reason Reasoning 4B — 4B image-text-to-text reasoning model, Apache-2.0. 4B params, Apache-2.0. Released Sep 3. 📦 Open weights · HF. → Source
📊 Surfaced 5 · scanned 9 today
🤖 ModelBytes Digest
Friday, September 04, 2026
Today’s split between fully open small models and a 438B reasoning model means builders can pick auditability or raw scale.
━━━ OPEN FRONTIER 🔓
K2 Horizon — Fully open release with weights, code, and training data, rare for builders who need to audit or reproduce the full stack. Released Sep 3. Open weights, code, training data. → Source
Spark X2.5 4B — A compact 4B text model under Apache 2.0, useful for lightweight generation without a closed API. Released Sep 4. 4B params, Apache 2.0. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Quasar 438B — A 438B reasoning model aimed at complex multi-step inference, pushing parameter scale for reasoning-heavy workloads. Released Sep 4. 438B params. → Source
━━━ SPECIALIZED 🎯
GaussianMind — Audio-text-to-text in 5.6B params with Apache 2.0, targeting speech-plus-language tasks without a closed API. Released Sep 3. 5.6B params, Apache 2.0. 📦 Open weights · HF. → Source
📊 Surfaced 4 · scanned 6 today
Friday, September 04, 2026
Today’s split between fully open small models and a 438B reasoning model means builders can pick auditability or raw scale.
━━━ OPEN FRONTIER 🔓
K2 Horizon — Fully open release with weights, code, and training data, rare for builders who need to audit or reproduce the full stack. Released Sep 3. Open weights, code, training data. → Source
Spark X2.5 4B — A compact 4B text model under Apache 2.0, useful for lightweight generation without a closed API. Released Sep 4. 4B params, Apache 2.0. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Quasar 438B — A 438B reasoning model aimed at complex multi-step inference, pushing parameter scale for reasoning-heavy workloads. Released Sep 4. 438B params. → Source
━━━ SPECIALIZED 🎯
GaussianMind — Audio-text-to-text in 5.6B params with Apache 2.0, targeting speech-plus-language tasks without a closed API. Released Sep 3. 5.6B params, Apache 2.0. 📦 Open weights · HF. → Source
📊 Surfaced 4 · scanned 6 today
🤖 ModelBytes Digest
Saturday, September 05, 2026
Today's releases extend frontier API access to long-context multimodal work, with OpenAI's Astra line and a fresh Qwen Max snapshot both targeting builders who need 1M+ token windows and vision input.
━━━ OPEN FRONTIER 🔓
Qwen3.8 Max 0902 — Updated snapshot of Alibaba's 2.4T-param MoE that takes text, image, and video input and returns text, aimed at builders needing a cheaper multimodal frontier alternative. Released Sep 3 · 1.0M context · $2.00/$6.00 per 1M. ⚡ API live · OpenRouter. → Source
━━━ CLOSED FRONTIER 🔒
GPT-6 Astra — OpenAI's flagship for end-to-end work: advanced analysis, software engineering, deep research, scientific work, and document creation, with a 1.1M context window. Released Sep 4 · 1.1M context · $10.00/$50.00 per 1M. ⚡ API live · OpenRouter. → Source
GPT-6 Astra Pro — Same underlying Astra model served with reasoning.mode=pro for higher-quality responses on complex tasks, for builders who want a toggle rather than a separate model. Released Sep 4 · 1.1M context · $10.00/$50.00 per 1M. ⚡ API live · OpenRouter. → Source
📊 Surfaced 3 · scanned 6 today
Saturday, September 05, 2026
Today's releases extend frontier API access to long-context multimodal work, with OpenAI's Astra line and a fresh Qwen Max snapshot both targeting builders who need 1M+ token windows and vision input.
━━━ OPEN FRONTIER 🔓
Qwen3.8 Max 0902 — Updated snapshot of Alibaba's 2.4T-param MoE that takes text, image, and video input and returns text, aimed at builders needing a cheaper multimodal frontier alternative. Released Sep 3 · 1.0M context · $2.00/$6.00 per 1M. ⚡ API live · OpenRouter. → Source
━━━ CLOSED FRONTIER 🔒
GPT-6 Astra — OpenAI's flagship for end-to-end work: advanced analysis, software engineering, deep research, scientific work, and document creation, with a 1.1M context window. Released Sep 4 · 1.1M context · $10.00/$50.00 per 1M. ⚡ API live · OpenRouter. → Source
GPT-6 Astra Pro — Same underlying Astra model served with reasoning.mode=pro for higher-quality responses on complex tasks, for builders who want a toggle rather than a separate model. Released Sep 4 · 1.1M context · $10.00/$50.00 per 1M. ⚡ API live · OpenRouter. → Source
📊 Surfaced 3 · scanned 6 today
🤖 ModelBytes Digest
Sunday, September 06, 2026
Today’s split: a 16B Apache-2.0 open-weight text model for self-hosting, while xAI pushes a managed coding/agent frontier model—builders can choose local control or closed capability.
━━━ OPEN FRONTIER 🔓
LLaDA 2.2 Mini — A 16.3B-param Apache-2.0 text-generation model, small enough to self-host while staying permissively licensed. Released Sep 5. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Grok 4.6 — xAI’s most capable model, built for coding, agentic tasks, and knowledge work. ⚡ API live · Google Cloud Agent Platform. → Source
📊 Surfaced 2 · scanned 2 today
Sunday, September 06, 2026
Today’s split: a 16B Apache-2.0 open-weight text model for self-hosting, while xAI pushes a managed coding/agent frontier model—builders can choose local control or closed capability.
━━━ OPEN FRONTIER 🔓
LLaDA 2.2 Mini — A 16.3B-param Apache-2.0 text-generation model, small enough to self-host while staying permissively licensed. Released Sep 5. 📦 Open weights · HF. → Source
━━━ CLOSED FRONTIER 🔒
Grok 4.6 — xAI’s most capable model, built for coding, agentic tasks, and knowledge work. ⚡ API live · Google Cloud Agent Platform. → Source
📊 Surfaced 2 · scanned 2 today