Home › Best › best long context open LLM The best long-context open LLMs (128K+ tokens) Open LLMs with a native context window of 128K tokens or more — for long documents, big codebases, and extended chats. Ranked by context window (longest first), with the honest cost to run each. Context length is a hard catalog number; how well a model actually uses it is yours to judge.
How this is ranked: Objective: context window is a catalog fact. We rank by the number itself and add the honest caveat that a large native window doesn't guarantee good long-context recall — we report the spec, not a benchmark we didn't run.
1. Llama 4 Scout (17B-16E) — Meta, 109B · 1M ctx · $0.20/1M API · commercial OK2. Llama 4 Maverick (17B-128E) — Meta, 402B · 1M ctx · $0.38/1M API · commercial OK3. DeepSeek V4 Pro — deepseek-ai, 861.6B · 1M ctx · $0.65/1M API · commercial OK4. DeepSeek V4 Flash — deepseek-ai, 158.1B · 1M ctx · $0.21/1M API · commercial OK5. MiMo V2.5 — XiaomiMiMo, 310.8B · 1M ctx · $2.59/1M API est. · commercial OK6. Vikhr Nemo 12B Instruct R 21 09 24 — Vikhrmodels, 12.2B · 1M ctx · $0.20/1M API est. · commercial OK7. DeepSeek V4 Flash 0731 — deepseek-ai, 304.2B · 1M ctx · $0.14/1M API · commercial OK8. DeepSeek V4 Flash NVFP4 — nvidia, 166.7B · 1M ctx · $1.43/1M API est. · commercial OK9. DeepSeek V4 Pro NVFP4 — nvidia, 910B · 1M ctx · $7.38/1M API est. · commercial OK10. GLM 5.2 NVFP4 — nvidia, 381B · 1M ctx · $3.15/1M API est. · commercial OK11. mistralai Mistral Nemo Instruct 2407 — SillyTilly, 12.2B · 1M ctx · $0.20/1M API est. · commercial OK12. Qwen2.5 7B Instruct 1M — Qwen, 7.6B · 1M ctx · $0.16/1M API est. · commercial OK13. DeepSeek V4 Flash DSpark — deepseek-ai, 165.3B · 1M ctx · $1.42/1M API est. · commercial OK14. Laguna S 2.1 — poolside, 117.6B · 1M ctx · $0.14/1M API · non-commercial15. MiMo V2.5 Pro — XiaomiMiMo, 1023.2B · 1M ctx · $8.29/1M API est. · commercial OK16. MiMo V2.5 Pro FP4 DFlash — XiaomiMiMo, 554.3B · 1M ctx · $4.53/1M API est. · commercial OK17. DeepSeek V4 Flash — unsloth, 158.1B · 1M ctx · $1.36/1M API est. · commercial OK18. GLM 5.2 NVFP4 — lukealonso, 432B · 1M ctx · $3.56/1M API est. · commercial OK19. GLM 5.2 — zai-org, 753.3B · 1M ctx · $1.47/1M API · commercial OK20. DeepSeek V4 Flash 0731 NVFP4 — MJPansa, 304.2B · 1M ctx · $2.53/1M API est. · commercial OK21. DeepSeek V4 Pro DSpark — deepseek-ai, 889.5B · 1M ctx · $7.22/1M API est. · commercial OK22. GLM 5.2 504B — 0xSero, 290B · 1M ctx · $2.42/1M API est. · commercial OK23. GLM 5.2 W4AFP8 — PhalaCloud, 391.9B · 1M ctx · $3.24/1M API est. · commercial OK24. MiniMax M3 EAGLE3 — Inferact, 3.3B · 1M ctx · $0.13/1M API est. · commercial OK25. GLM 5.2 NVFP4 — Mapika, 379.2B · 1M ctx · $3.13/1M API est. · commercial OK26. Seed OSS 36B Instruct — ByteDance-Seed, 36.2B · 524.3K ctx · $0.39/1M API est. · commercial OK27. Hermes 4.3 36B — NousResearch, 36.2B · 524.3K ctx · $0.39/1M API est. · commercial OK28. ReaderLM v2 — jinaai, 1.5B · 512.8K ctx · $0.11/1M API est. · non-commercial29. North Mini Code 1.0 — CohereLabs, 30.5B · 500K ctx · $0.34/1M API est. · commercial OK30. LongCat Flash Lite — meituan-longcat, 69.1B · 327.7K ctx · $0.65/1M API est. · commercial OK31. Qwen3 4B Instruct 2507 — Qwen, 4B · 262.1K ctx · $0.13/1M API est. · commercial OK32. Rio 3.0 Open Mini — prefeitura-rio, 4B · 262.1K ctx · $0.13/1M API est. · commercial OK33. Qwen3 Coder 30B A3B Instruct — Qwen, 30.5B · 262.1K ctx · $0.18/1M API · commercial OK34. OTel 2.0 LLM 31B IT — farbodtavakkoli, 32.1B · 262.1K ctx · $0.36/1M API est. · commercial OK35. Qwen3 Coder Next — Qwen, 79.7B · 262.1K ctx · $0.46/1M API · commercial OK36. Qwen3 30B A3B Instruct 2507 — Qwen, 30.5B · 262.1K ctx · $0.12/1M API · commercial OK37. Qwen3 4B Thinking 2507 — Qwen, 4B · 262.1K ctx · $0.13/1M API est. · commercial OK38. Step 3.5 Flash — stepfun-ai, 199.4B · 262.1K ctx · $1.70/1M API est. · commercial OK39. Qwen3 Next 80B A3B Instruct — Qwen, 81.3B · 262.1K ctx · $0.60/1M API · commercial OK40. Laguna XS.2 — poolside, 33.4B · 262.1K ctx · $0.15/1M API · commercial OKShowing the top 40 of 172. See all →
More: all "best" lists · Outcome Lab · all models
Open the free Spanvero advisor → · Honest, $0-markup. © 2026 Cynosure LLC.