/best/researchupdated July 2026120 models evaluated
Best AI for research
Synthesizing sources, answering with citations, deep reading.
Research workloads punish hallucination harder than almost any other task. The model has to read long, contradictory sources and produce an answer that a specialist wouldn't laugh at. Long-context retention and tool-use (search, browsing) are table stakes — a model with a 128K context window that loses track of page 1 by page 100 is useless for real research.
What matters: long-context faithfulness, reasoning chain quality, and willingness to say 'I don't know' rather than confabulate. Models with retrieval-augmented architectures (like Sonar Pro) have a natural edge here, but frontier chat models with large context windows are competitive when source material is provided directly.
Our research rank weights long-context (2×), reasoning (1.5×), tool-use (1×), and multilingual (0.5×). Multilingual matters more than people expect — research sources are rarely all in English.
Full rankings
All 120 models, scored for research
| # | Model | Provider | Task score | $/in | $/out | Context |
|---|---|---|---|---|---|---|
| 01 | Qwen: Qwen3 235B A22B Instruct 2507 | QQwen | 5.00/5.0 | $0.090 | $0.100 | 262K |
| 02 | OpenAI: gpt-oss-120b | OOpenAI | 5.00/5.0 | $0.036 | $0.180 | 131K |
| 03 | NVIDIA: Nemotron 3 Nano 30B A3B | NNVIDIA | 5.00/5.0 | $0.050 | $0.200 | 262K |
| 04 | MiMo-V2.5 | XXiaomi | 5.00/5.0 | $0.105 | $0.280 | 1.0M |
| 05 | Xiaomi: MiMo-V2-Flash | XXiaomi | 5.00/5.0 | $0.100 | $0.300 | 262K |
| 06 | StepFun: Step 3.5 Flash | SStepfun | 5.00/5.0 | $0.100 | $0.300 | 262K |
| 07 | Gemma 4 26B A4B | GGoogle | 5.00/5.0 | $0.060 | $0.330 | 262K |
| 08 | DeepSeek V3.2 | DDeepSeek | 5.00/5.0 | $0.214 | $0.322 | 131K |
| 09 | GLM-4.7 Flash | ZZhipu AI | 5.00/5.0 | $0.060 | $0.400 | 203K |
| 10 | NVIDIA: Nemotron 3 Super | NNVIDIA | 5.00/5.0 | $0.080 | $0.450 | 1M |
| 11 | Grok 4.1 Fast | XXAI | 5.00/5.0 | $0.200 | $0.500 | 2M |
| 12 | Inception: Mercury 2 | IInception | 5.00/5.0 | $0.250 | $0.750 | 128K |
| 13 | MiMo-V2.5-Pro | XXiaomi | 5.00/5.0 | $0.435 | $0.870 | 1.0M |
| 14 | Qwen: Qwen3.6 35B A3B | QQwen | 5.00/5.0 | $0.140 | $1.00 | 262K |
| 15 | Qwen: Qwen3.5-35B-A3B | QQwen | 5.00/5.0 | $0.140 | $1.00 | 262K |
| 16 | MiniMax: MiniMax M3 | MMinimax | 5.00/5.0 | $0.300 | $1.20 | 1.0M |
| 17 | Qwen: Qwen3.7 Plus | QQwen | 5.00/5.0 | $0.320 | $1.28 | 1M |
| 18 | Z.ai: GLM 5.2 | ZZhipu AI | 5.00/5.0 | $0.406 | $1.28 | 1.0M |
| 19 | GLM-4.7 | ZZhipu AI | 5.00/5.0 | $0.400 | $1.75 | 203K |
| 20 | Qwen: Qwen3.5 Plus 2026-04-20 | QQwen | 5.00/5.0 | $0.300 | $1.80 | 1M |
| 21 | Qwen: Qwen3.6 Plus | QQwen | 5.00/5.0 | $0.325 | $1.95 | 1M |
| 22 | GPT-5 Mini | OOpenAI | 5.00/5.0 | $0.250 | $2.00 | 400K |
| 23 | ByteDance Seed: Seed 1.6 | BBytedance-seed | 5.00/5.0 | $0.250 | $2.00 | 262K |
| 24 | ByteDance Seed: Seed-2.0-Lite | BBytedance-seed | 5.00/5.0 | $0.250 | $2.00 | 262K |
| 25 | xAI: Grok Build 0.1 | XXAI | 5.00/5.0 | $1.00 | $2.00 | 256K |
| 26 | NVIDIA: Nemotron 3 Ultra | NNVIDIA | 5.00/5.0 | $0.500 | $2.20 | 1M |
| 27 | Qwen: Qwen3.6 27B | QQwen | 5.00/5.0 | $0.285 | $2.40 | 262K |
| 28 | Grok 4.20 | XXAI | 5.00/5.0 | $1.25 | $2.50 | 2M |
| 29 | Gemini 3 Flash Preview | GGoogle | 5.00/5.0 | $0.500 | $3.00 | 1.0M |
| 30 | MoonshotAI: Kimi K2.6 | MMoonshotAI | 5.00/5.0 | $0.660 | $3.41 | 262K |
| 31 | Qwen 3.7 Max | QQwen | 5.00/5.0 | $1.25 | $3.75 | 1M |
| 32 | GPT-5.4 Mini | OOpenAI | 5.00/5.0 | $0.750 | $4.50 | 400K |
| 33 | o4 Mini | OOpenAI | 5.00/5.0 | $1.10 | $4.40 | 200K |
| 34 | Claude Haiku 4.5 | AAnthropic | 5.00/5.0 | $1.00 | $5.00 | 200K |
| 35 | OpenAI: GPT-5.6 Luna | OOpenAI | 5.00/5.0 | $1.00 | $6.00 | 1.1M |
| 36 | OpenAI: GPT-5.6 Luna Pro | OOpenAI | 5.00/5.0 | $1.00 | $6.00 | 1.1M |
| 37 | Qwen: Qwen3.6 Max Preview | QQwen | 5.00/5.0 | $1.04 | $6.24 | 262K |
| 38 | Grok 4.5 | XXAI | 5.00/5.0 | $2.00 | $6.00 | 500K |
| 39 | GPT-4.1 | OOpenAI | 5.00/5.0 | $2.00 | $8.00 | 1.0M |
| 40 | GPT-5 | OOpenAI | 5.00/5.0 | $1.25 | $10.00 | 400K |
| 41 | GPT-5.1 | OOpenAI | 5.00/5.0 | $1.25 | $10.00 | 400K |
| 42 | Gemini 3.1 Pro Preview | GGoogle | 5.00/5.0 | $2.00 | $12.00 | 1.0M |
| 43 | GPT-5.2 | OOpenAI | 5.00/5.0 | $1.75 | $14.00 | 400K |
| 44 | GPT-5.4 | OOpenAI | 5.00/5.0 | $2.50 | $15.00 | 1.1M |
| 45 | OpenAI: GPT-5.6 Terra | OOpenAI | 5.00/5.0 | $2.50 | $15.00 | 1.1M |
| 46 | OpenAI: GPT-5.6 Terra Pro | OOpenAI | 5.00/5.0 | $2.50 | $15.00 | 1.1M |
| 47 | Claude Sonnet 4.6 | AAnthropic | 5.00/5.0 | $3.00 | $15.00 | 1M |
| 48 | Grok 4 | XXAI | 5.00/5.0 | $3.00 | $15.00 | 256K |
| 49 | Grok 3 | XXAI | 5.00/5.0 | $3.00 | $15.00 | 131K |
| 50 | Claude Opus 4.7 | AAnthropic | 5.00/5.0 | $5.00 | $25.00 | 1M |
| 51 | Claude Opus 4.6 | AAnthropic | 5.00/5.0 | $5.00 | $25.00 | 1M |
| 52 | OpenAI: GPT-5.6 Sol | OOpenAI | 5.00/5.0 | $5.00 | $30.00 | 1.1M |
| 53 | Anthropic: Claude Opus 4.8 (Fast) | AAnthropic | 5.00/5.0 | $10.00 | $50.00 | 1M |
| 54 | Anthropic: Claude Fable 5 | AAnthropic | 5.00/5.0 | $10.00 | $50.00 | 1M |
| 55 | Poolside: Laguna XS 2.1 | PPoolside | 4.67/5.0 | $0.060 | $0.120 | 262K |
| 56 | DeepSeek V4 Flash | DDeepSeek | 4.67/5.0 | $0.077 | $0.154 | 1.0M |
| 57 | Gemma 4 31B | GGoogle | 4.67/5.0 | $0.120 | $0.350 | 262K |
| 58 | inclusionAI: Ring-2.6-1T | IInclusionai | 4.67/5.0 | $0.075 | $0.625 | 262K |
| 59 | inclusionAI: Ling-2.6-1T | IInclusionai | 4.67/5.0 | $0.075 | $0.625 | 262K |
| 60 | DeepSeek V3.1 | DDeepSeek | 4.67/5.0 | $0.210 | $0.790 | 164K |
| 61 | Arcee AI: Trinity Large Thinking | AArcee-ai | 4.67/5.0 | $0.250 | $0.800 | 262K |
| 62 | DeepSeek V4 Pro | DDeepSeek | 4.67/5.0 | $0.435 | $0.870 | 1.0M |
| 63 | MiniMax: MiniMax M2.7 | MMinimax | 4.67/5.0 | $0.240 | $0.960 | 205K |
| 64 | Nex-N2-Pro | NNex AGI | 4.67/5.0 | $0.250 | $1.00 | 262K |
| 65 | MiniMax: MiniMax M2 | MMinimax | 4.67/5.0 | $0.255 | $1.02 | 205K |
| 66 | Qwen: Qwen3.6 Flash | QQwen | 4.67/5.0 | $0.188 | $1.13 | 1M |
| 67 | StepFun: Step 3.7 Flash | SStepfun | 4.67/5.0 | $0.200 | $1.15 | 256K |
| 68 | Kwaipilot: KAT-Coder-Pro V2 | KKwaipilot | 4.67/5.0 | $0.300 | $1.20 | 256K |
| 69 | GPT-5.4 Nano | OOpenAI | 4.67/5.0 | $0.200 | $1.25 | 400K |
| 70 | Gemini 3.1 Flash Lite Preview | GGoogle | 4.67/5.0 | $0.250 | $1.50 | 1.0M |
| 71 | Mistral Medium 3.1 | MMistral | 4.67/5.0 | $0.400 | $2.00 | 131K |
| 72 | R1 0528 | DDeepSeek | 4.67/5.0 | $0.500 | $2.15 | 164K |
| 73 | R1 | DDeepSeek | 4.67/5.0 | $0.700 | $2.50 | 164K |
| 74 | Grok 4.3 | XXAI | 4.67/5.0 | $1.25 | $2.50 | 1M |
| 75 | Kimi K2.7 Code | MMoonshot AI | 4.67/5.0 | $0.720 | $3.50 | 262K |
| 76 | Mistral Medium 3.5 | MMistral | 4.67/5.0 | $1.50 | $7.50 | 262K |
| 77 | o3 | OOpenAI | 4.67/5.0 | $2.00 | $8.00 | 200K |
| 78 | Gemini 3.5 Flash | GGoogle | 4.67/5.0 | $1.50 | $9.00 | 1.0M |
| 79 | Gemini 2.5 Pro | GGoogle | 4.67/5.0 | $1.25 | $10.00 | 1.0M |
| 80 | Anthropic: Claude Sonnet 5 | AAnthropic | 4.67/5.0 | $2.00 | $10.00 | 1M |
| 81 | Anthropic: Claude Opus 4.8 | AAnthropic | 4.67/5.0 | $5.00 | $25.00 | 1M |
| 82 | GPT-5.5 | OOpenAI | 4.67/5.0 | $5.00 | $30.00 | 1.1M |
| 83 | OpenAI: GPT-5.6 Sol Pro | OOpenAI | 4.67/5.0 | $5.00 | $30.00 | 1.1M |
| 84 | inclusionAI: Ling-2.6-flash | IInclusionai | 4.33/5.0 | $0.010 | $0.030 | 262K |
| 85 | IBM: Granite 4.1 8B | IIbm-granite | 4.33/5.0 | $0.050 | $0.100 | 131K |
| 86 | OpenAI: gpt-oss-20b | OOpenAI | 4.33/5.0 | $0.030 | $0.140 | 131K |
| 87 | Qwen: Qwen3.5-9B | QQwen | 4.33/5.0 | $0.100 | $0.150 | 262K |
| 88 | Qwen: Qwen3 30B A3B Instruct 2507 | QQwen | 4.33/5.0 | $0.048 | $0.193 | 131K |
| 89 | GPT-5 Nano | OOpenAI | 4.33/5.0 | $0.050 | $0.400 | 400K |
| 90 | ByteDance Seed: Seed-2.0-Mini | BBytedance-seed | 4.33/5.0 | $0.100 | $0.400 | 262K |
| 91 | Gemini 2.5 Flash Lite | GGoogle | 4.33/5.0 | $0.100 | $0.400 | 1.0M |
| 92 | Grok 3 Mini | XXAI | 4.33/5.0 | $0.300 | $0.500 | 131K |
| 93 | MiniMax: MiniMax M2.5 | MMinimax | 4.33/5.0 | $0.150 | $0.900 | 205K |
| 94 | DeepSeek V3.1 Terminus | DDeepSeek | 4.33/5.0 | $0.270 | $0.950 | 164K |
| 95 | MiniMax: MiniMax M2.1 | MMinimax | 4.33/5.0 | $0.300 | $1.20 | 205K |
| 96 | Google: Gemini 3.1 Flash Lite | GGoogle | 4.33/5.0 | $0.250 | $1.50 | 1.0M |
| 97 | Mistral Large 3 2512 | MMistral | 4.33/5.0 | $0.500 | $1.50 | 262K |
| 98 | GPT-4.1 Mini | OOpenAI | 4.33/5.0 | $0.400 | $1.60 | 1.0M |
| 99 | Devstral 2 2512 | MMistral | 4.33/5.0 | $0.400 | $2.00 | 262K |
| 100 | MiniMax: MiniMax M1 | MMinimax | 4.33/5.0 | $0.400 | $2.20 | 1M |
| 101 | Ministral 3 14B 2512 | MMistral | 4.00/5.0 | $0.200 | $0.200 | 262K |
| 102 | ByteDance Seed: Seed 1.6 Flash | BBytedance-seed | 4.00/5.0 | $0.075 | $0.300 | 262K |
| 103 | Llama 3.3 70B Instruct | MMeta | 4.00/5.0 | $0.100 | $0.320 | 131K |
| 104 | Mistral Small 4 | MMistral | 4.00/5.0 | $0.150 | $0.600 | 262K |
| 105 | Mistral Small 3.1 24B | MMistral | 4.00/5.0 | $0.351 | $0.555 | 128K |
| 106 | Codestral 2508 | MMistral | 4.00/5.0 | $0.300 | $0.900 | 256K |
| 107 | MiniMax: MiniMax-01 | MMinimax | 4.00/5.0 | $0.200 | $1.10 | 1.0M |
| 108 | Gemini 2.5 Flash | GGoogle | 4.00/5.0 | $0.300 | $2.50 | 1.0M |
| 109 | Ministral 3 3B 2512 | MMistral | 3.67/5.0 | $0.100 | $0.100 | 131K |
| 110 | Ministral 3 8B 2512 | MMistral | 3.67/5.0 | $0.150 | $0.150 | 262K |
| 111 | Qwen: Qwen3 Coder 30B A3B Instruct | QQwen | 3.67/5.0 | $0.070 | $0.270 | 160K |
| 112 | Llama 4 Scout | MMeta | 3.67/5.0 | $0.100 | $0.300 | 10M |
| 113 | GPT-4.1 Nano | OOpenAI | 3.67/5.0 | $0.100 | $0.400 | 1.0M |
| 114 | Grok Code Fast 1 | XXAI | 3.67/5.0 | $0.200 | $1.50 | 256K |
| 115 | Mistral Small 3.2 24B | MMistral | 3.33/5.0 | $0.075 | $0.200 | 128K |
| 116 | Llama 4 Maverick | MMeta | 3.33/5.0 | $0.150 | $0.600 | 1.0M |
| 117 | MiniMax: MiniMax M2-her | MMinimax | 3.33/5.0 | $0.300 | $1.20 | 66K |
| 118 | GPT-4o | OOpenAI | 3.33/5.0 | $2.50 | $10.00 | 128K |
| 119 | IBM: Granite 4.0 Micro | IIbm-granite | 3.00/5.0 | $0.017 | $0.112 | 131K |
| 120 | GPT-4o-mini | OOpenAI | 3.00/5.0 | $0.150 | $0.600 | 128K |
Pricing — top 5 for research
Q
5.00/5.0
O
5.00/5.0
N
5.00/5.0
X
5.00/5.0
X
5.00/5.0
The best AI for research changes every month.
We'll email you when rankings shift, new models hit the top 5, or pricing cuts reshuffle the value leaders.
Get notified when models change
Price drops, new models, benchmark updates. One email per change, no spam.