1stTop pick
Researched 4d agoArrow 2 Telos
- Context
- 1.05m
- Output (from)
- $30.00 / 1M
Best LLMs for Retrieval-Augmented Generation (2026)
Last refreshed 2026-09-21. Next refresh: weekly.
Best LLMs for RAG in 2026 — ranked by context window, retrieval benchmarks, and tool support. Covers document QA to enterprise search.
Verdict
Use Arrow 2 Telos for RAG today.
GPT-6 Astra is the runner-up: 1.05m vs 1.05m on Context.
2ndShortlistRestricted
Researched 18d agoGPT-6 Astra
- Context
- 1.05m
- Output (from)
- $50.00 / 1M
3rdShortlist
Researched todayXiaomi MiMo-V2.6-Pro
- Context
- 1.05m
- Output (from)
- $0.870 / 1M
How we rank
RAG picks emphasize the strongest sourced long-document benchmark among tracked RAG needles, QA, and retrieval suites, then context window, then recency.
- Eligibility — Models tagged for the RAG decision task (collections/specialization fit or scores on RULER, ZeroSCROLLs, InfiniteBench, multi-needle, MS MARCO, SQuAD, NaturalQuestions, TriviaQA, etc.).
- Primary ranking — Best score across the RAG benchmark bundle, then larger declared context window, then newer release.
- Variant collapse — We keep one row per model family (`familySlug` + parameter tier). When headline scores tie within ±0.5 pt (±10 Elo on Chatbot Arena), we pick the canonical SKU by lowest tracked input price, then GA over preview or limited access, then newest `release`. A folded sibling within the benchmark noise band can show a "Tied within margin" chip on that score cell.
- Pricing — Lowest tracked provider input/output where present.
| # | Model | Input $/1M | Output $/1M | |
|---|---|---|---|---|
| 1 | Nemotron 3 Super-120B-A12B Signal used: RULER 96.33% | $0.09 | $0.45 | |
| 2 | Llama 4 Scout 17B-16E Instruct Vision Signal used: Context 10m | $0.08 | $0.22 | |
| 3 | Gemini 1.5 Pro Signal used: Context 2m | $1.25 | $5.00 | |
| 4 | Arrow 2 Telos VisionTools Signal used: Context 1.05m | $6.00 | $30.00 | |
| 5 | GPT-6 Astra RestrictedReasoningVisionTools Signal used: Context 1.05m | $10.00 | $50.00 | |
| 6 | GPT-5.6 Luna ReasoningVisionTools Signal used: Context 1.05m | $1.00 | $6.00 | |
| 7 | GPT-5.6 Sol ReasoningVisionTools Signal used: Context 1.05m | $5.00 | $30.00 | |
| 8 | GPT-5.6 Terra ReasoningVisionTools Signal used: Context 1.05m | $2.50 | $15.00 | |
| 9 | GPT-5.5 ReasoningVisionTools Signal used: Context 1.05m | $5.00 | $30.00 | |
| 10 | GPT-5.5 Pro ReasoningVisionTools Signal used: Context 1.05m | $30.00 | $180.00 | |
| 11 | GPT-5.4 ReasoningVisionTools Signal used: Context 1.05m | $2.50 | $15.00 | |
| 12 | GPT-5.4 Pro ReasoningVisionTools Signal used: Context 1.05m | $30.00 | $180.00 | |
| 13 | Xiaomi MiMo-V2.6-Pro ReasoningVisionTools Signal used: Context 1.05m | $0.43 | $0.87 | |
| 14 | Nex-N2.5-Max ReasoningTools Signal used: Context 1.05m | — | — | |
| 15 | Gemini 3.8 Flash ReasoningVisionTools Signal used: Context 1.05m | $0.75 | $3.75 | |
| 16 | Muse Spark 1.3 ReasoningVisionTools Signal used: Context 1.05m | $1.25 | $4.25 | |
| 17 | Muse Spark 1.3 Contributor ReasoningVisionTools Signal used: Context 1.05m | $0.10 | $0.20 | |
| 18 | Spark X2.5 1.7B ReasoningTools Signal used: Context 1.05m | — | — | |
| 19 | Spark X2.5 4B ReasoningTools Signal used: Context 1.05m | — | — | |
| 20 | Hunyuan Hy4 Preview PreviewReasoningTools Signal used: Context 1.05m | $0.83 | $2.50 |
Honorable mentions
Next seats in this ranking. Lines below are from each model's stored description in LLMReference seed data—spot-check the model page before relying on a capability claim.
Nex-N2.5-Max is Nex AGI's trillion-scale text-only MoE agentic model in the Nex-N2.5 family, released September 8, 2026. Apache-2 weights on Hugging Face (nex-agi/Nex-N2.5-Max). 1.6T total / 49B active MoE with a native 1,048,576-token context window (config max_position_embeddings). Supports reasoning, function calling, and tool use; no vision or multimodal inputs. No OpenRouter listing in this seed pass. Sibling Nex-N2.5-mini is the multimodal MoE variant; Nex-N2.5-Pro is held from this seed pass.
1.05m
Context
Gemini 3.8 Flash is Google DeepMind's generally available most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows, released September 2, 2026. It accepts text, image, video, audio, and PDF inputs and returns text, with a 1,048,576-token context window, up to 65,536 output tokens, and Gemini API support for thinking (low/medium/high; minimal is not supported), function calling, tool use, structured outputs, code execution, prompt caching, search grounding, URL context, computer use (preview), and batch, flex, and priority consumption. Official model ID: gemini-3.8-flash. Sibling of live gemini-3.7-flash under a new family gemini-3.8. Compare it for Coding, Agents, Long context, Vision, and JSON / Tool use.
1.05m
Context
Muse Spark 1.3 is Meta Superintelligence Labs' latest Muse Spark checkpoint, released September 2, 2026. Official Meta Model API model ID: muse-spark-1.3. First-party models docs: the latest version, recommended for new work, tuned for agentic workflows (multi-step tool, browser, and long-horizon tasks) with improved coding over 1.2; default model in docs examples. Product line (dev.meta.ai): Built for long-horizon coding and agentic work. Cleaner code, fewer tokens. First-party research post: rolling out today in Muse Code and Meta Model API; previously available reasoning modes available today with max reasoning coming shortly after additional safety testing. Official table: text, image, video, audio*, PDF input and text output; 1,048,576-token context. Official quickstart limits: context 1048576, output 131072; input modalities text/image/pdf/video, output text. Audio understanding in 1.3 is currently not fully supported (use 1.2 or Muse Voice Transcribe for audio). Tool/function calling, structured outputs, prompt caching, search grounding, and reasoning_effort (minimal/low/medium/high/xhigh; none is not supported). Compare it for Coding, Agents, Long context, and Vision.
1.05m
Context
Compare Top Picks
Side-by-side comparison of the top picks by price, benchmark, and API access.
Browse Other Categories
Best LLMs for Code GenerationBest LLMs for Agents vs Agent Products (2026)Best LLMs for ClassificationBest Open Source LLMsBest Multimodal / Vision LLMsBest LLM for Translation in 2026Best AI Image Models in 2026Best AI Video Models in 2026Best LLMs for Reasoning & MathBest Small Language Models (SLMs)Best LLMs for JSON / Tool useCheapest LLM APIs You Can Call Right NowBest Long Context LLMsBest Mainstream LLM APIs, RankedBest LLMs for EnterpriseBest Free LLMs You Can Use Right NowBest LLMs for WritingBest LLMs for MarketingBest LLMs for Customer Support