The AI Value Leaderboard: Intelligence per Dollar
Benchmarks tell you which model is smartest. Prices tell you which is cheapest. This table answers the question that matters: which model gives you the most capability for your money?
| Value rank | Model | Lab | Value Score | Arena Score | Arena rank | Blended $/1M |
|---|---|---|---|---|---|---|
| #1 | gpt-oss-20b | OpenAI | 2425 | 1287 ±6 | #108 | $0.04 |
| #2 | gpt-oss-120b | OpenAI | 2361.6 | 1366 ±4 | #86 | $0.07 |
| #3 | Qwen3 30B A3B Instruct 2507open | Qwen | 2174.7 | 1384 ±5 | #76 | $0.08 |
| #4 | Gemma 4 26B A4B | 1936 | 1435 ±8 | #44 | $0.12 | |
| #5 | Gemma 3 12B | 1788 | 1334 ±9 | #98 | $0.08 | |
| #6 | Qwen3.5-Flashopen | Qwen | 1738 | 1398 ±4 | #70 | $0.11 |
| #7 | Gemma 4 31B | 1584.9 | 1442 ±8 | #36 | $0.15 | |
| #8 | Gemma 3 4B | 1451.2 | 1291 ±9 | #107 | $0.06 | |
| #9 | Gemma 3n 4B | 1404 | 1305 ±5 | #105 | $0.08 | |
| #10 | Qwen3 32Bopen | Qwen | 1076.9 | 1340 ±9 | #96 | $0.13 |
| #11 | Gemini 2.5 Flash Lite Preview 09-2025 | 1025.1 | 1379 ±3 | #78 | $0.18 | |
| #12 | Gemini 2.5 Flash Lite | 966.9 | 1369 ±4 | #83 | $0.18 | |
| #13 | DeepSeek V4 Pro 0423open | DeepSeek | 960.2 | 1451 ±4 | #27 | $0.26 |
| #14 | Gemma 3 27B | 914.2 | 1358 ±4 | #90 | $0.17 | |
| #15 | GPT-5 Nano | OpenAI | 870.5 | 1320 ±7 | #102 | $0.14 |
| #16 | DeepSeek V3.2open | DeepSeek | 751.2 | 1425 ±4 | #51 | $0.3 |
| #17 | DeepSeek V4 Flash 0423open | DeepSeek | 676.8 | 1432 ±4 | #46 | $0.34 |
| #18 | Mistral Small 3open | Mistral | 582.6 | 1234 ±6 | #120 | $0.06 |
| #19 | MiniMax M2.7open | MiniMax | 556.7 | 1405 ±4 | #67 | $0.37 |
| #20 | Qwen3 30B A3Bopen | Qwen | 544.2 | 1317 ±5 | #104 | $0.22 |
| #21 | Qwen3.5-35B-A3Bopen | Qwen | 539.3 | 1396 ±4 | #73 | $0.36 |
| #22 | Llama 4 Scoutopen | Meta | 528.7 | 1279 ±5 | #113 | $0.15 |
| #23 | GPT-5.6 Luna | OpenAI | 510.9 | 1430 ±5 | #48 | $0.45 |
| #24 | GPT-4.1 Nano | OpenAI | 484 | 1285 ±8 | #111 | $0.18 |
| #25 | DeepSeek V3.1 Terminusopen | DeepSeek | 479.3 | 1417 ±10 | #59 | $0.45 |
| #26 | Llama 3.3 70B Instructopen | Meta | 477.4 | 1274 ±4 | #115 | $0.16 |
| #27 | Qwen3.7 Plusopen | Qwen | 453.9 | 1454 ±4 | #23 | $0.56 |
| #28 | Qwen3 Next 80B A3B Instructopen | Qwen | 446.2 | 1418 ±5 | #57 | $0.49 |
| #29 | MiniMax M3open | MiniMax | 444.8 | 1434 ±4 | #45 | $0.52 |
| #30 | DeepSeek V3.2 Expopen | DeepSeek | 434.8 | 1333 ±5 | #99 | $0.31 |
| #31 | Qwen3 Next 80B A3B Thinkingopen | Qwen | 406.8 | 1368 ±6 | #84 | $0.41 |
| #32 | Gemini 3.1 Flash Lite | 382.8 | 1415 ±4 | #60 | $0.56 | |
| #33 | GPT-5.4 Nano | OpenAI | 373.8 | 1373 ±4 | #80 | $0.46 |
| #34 | MiniMax M2.1open | MiniMax | 364.4 | 1391 ±5 | #74 | $0.52 |
| #35 | Qwen3 VL 235B A22B Instructopen | Qwen | 348.8 | 1421 ±7 | #54 | $0.63 |
| #36 | MiniMax M2.5open | MiniMax | 336.3 | 1359 ±4 | #89 | $0.47 |
| #37 | Qwen3 Coder 480B A35Bopen | Qwen | 329.5 | 1357 ±5 | #91 | $0.48 |
| #38 | GPT-4o-mini (2024-07-18) | OpenAI | 328.8 | 1286 ±4 | #110 | $0.26 |
| #39 | Qwen-Plusopen | Qwen | 323.8 | 1326 ±8 | #101 | $0.39 |
| #40 | Qwen3.6 Plusopen | Qwen | 323.7 | 1437 ±4 | #42 | $0.73 |
| #41 | Qwen3.5-122B-A10Bopen | Qwen | 304.8 | 1418 ±4 | #56 | $0.72 |
| #42 | Qwen3 235B A22B Thinking 2507open | Qwen | 287.8 | 1415 ±7 | #61 | $0.75 |
| #43 | Llama 4 Maverickopen | Meta | 287.1 | 1287 ±4 | #109 | $0.3 |
| #44 | Gemini 3.5 Flash Lite | 277.1 | 1436 ±5 | #43 | $0.85 | |
| #45 | Kimi K2.5open | Moonshot AI | 272.9 | 1446 ±3 | #33 | $0.9 |
| #46 | MiniMax M2open | MiniMax | 270.7 | 1342 ±8 | #94 | $0.52 |
| #47 | Kimi K2.6open | Moonshot AI | 265.2 | 1455 ±5 | #22 | $0.96 |
| #48 | Gemini 2.5 Flash | 255.6 | 1417 ±2 | #58 | $0.85 | |
| #49 | GPT-5 Mini | OpenAI | 251.6 | 1373 ±5 | #79 | $0.69 |
| #50 | Qwen3.8 27Bopen | Qwen | 250.2 | 1439 ±6 | #37 | $0.96 |
| #51 | R1 0528open | DeepSeek | 249.3 | 1428 ±6 | #49 | $0.91 |
| #52 | Qwen3.5-27Bopen | Qwen | 246.3 | 1408 ±4 | #65 | $0.85 |
| #53 | Gemini 3 Flash Preview | 237.1 | 1467 ±4 | #17 | $1.13 | |
| #54 | Qwen3.5 397B A17Bopen | Qwen | 219.1 | 1438 ±3 | #40 | $1.09 |
| #55 | Qwen3 235B A22Bopen | Qwen | 208.6 | 1366 ±5 | #85 | $0.8 |
| #56 | GPT-4.1 Mini | OpenAI | 200.3 | 1340 ±4 | #95 | $0.7 |
| #57 | Kimi K2 Thinkingopen | Moonshot AI | 199.5 | 1415 ±3 | #62 | $1.08 |
| #58 | Gemini 3.8 Flash | 196.5 | 1495 ±9 | #4 | $1.50 | |
| #59 | Gemini 3.7 Flash | 193.7 | 1491 ±8 | #6 | $1.50 | |
| #60 | Mistral Small 3.1 24Bopen | Mistral | 192.5 | 1277 ±5 | #114 | $0.4 |
| #61 | Nova 2 Lite | Amazon | 191.3 | 1363 ±6 | #88 | $0.85 |
| #62 | Phi 4 | Microsoft | 189.7 | 1217 ±5 | #123 | $0.09 |
| #63 | Gemini 3.6 Flash | 184.1 | 1476 ±5 | #14 | $1.50 | |
| #64 | Kimi K2 0711open | Moonshot AI | 170.4 | 1371 ±5 | #82 | $1.00 |
| #65 | Kimi K2 0905open | Moonshot AI | 167 | 1380 ±7 | #77 | $1.08 |
| #66 | Grok 4.20 | xAI | 160.4 | 1451 ±4 | #26 | $1.56 |
| #67 | Grok 4.20 Multi-Agent | xAI | 159.9 | 1450 ±4 | #30 | $1.56 |
| #68 | Qwen3 VL 235B A22B Thinkingopen | Qwen | 154.5 | 1401 ±7 | #69 | $1.30 |
| #69 | Qwen3 Maxopen | Qwen | 153.3 | 1439 ±5 | #38 | $1.56 |
| #70 | Llama 3.1 70B Instructopen | Meta | 152 | 1261 ±4 | #119 | $0.4 |
| #71 | R1open | DeepSeek | 150.1 | 1373 ±5 | #81 | $1.15 |
| #72 | MiniMax M1open | MiniMax | 148.2 | 1343 ±4 | #93 | $0.96 |
| #73 | Grok 4.3 | xAI | 126.5 | 1398 ±4 | #71 | $1.56 |
| #74 | GPT-5.4 Mini | OpenAI | 125.7 | 1412 ±4 | #64 | $1.69 |
| #75 | Qwen3.7 Maxopen | Qwen | 123.8 | 1474 ±10 | #15 | $2.21 |
| #76 | Qwen3.6 Max Previewopen | Qwen | 106.6 | 1446 ±8 | #31 | $2.31 |
| #77 | Claude Haiku 4.5 | Anthropic | 98.3 | 1397 ±3 | #72 | $2.00 |
| #78 | Qwen3.8 Maxopen | Qwen | 93.5 | 1481 ±6 | #9 | $3.00 |
| #79 | Gemini 3.5 Flash | 83.6 | 1482 ±4 | #8 | $3.38 | |
| #80 | Grok 4.5 | xAI | 83.4 | 1450 ±5 | #29 | $3.00 |
| #81 | o4 Mini | OpenAI | 79.6 | 1353 ±4 | #92 | $1.93 |
| #82 | Grok 4.6 | xAI | 76.6 | 1430 ±6 | #47 | $3.00 |
| #83 | Mistral Largeopen | Mistral | 75.7 | 1427 ±3 | #50 | $3.00 |
| #84 | Gemini 2.5 Pro | 75 | 1458 ±2 | #20 | $3.44 | |
| #85 | Mistral Medium 3.5open | Mistral | 73.5 | 1421 ±7 | #53 | $3.00 |
| #86 | Kimi K3open | Moonshot AI | 66 | 1472 ±5 | #16 | $4.13 |
| #87 | GPT-5.1 | OpenAI | 64.8 | 1423 ±4 | #52 | $3.44 |
| #88 | GPT-5.6 Sol | OpenAI | 63.8 | 1455 ±5 | #21 | $4.00 |
| #89 | Gemini 3.1 Pro Preview Custom Tools | 62.2 | 1480 ±3 | #10 | $4.50 | |
| #90 | Gemini 3.1 Pro Preview | 62.2 | 1480 ±3 | #11 | $4.50 | |
| #91 | Nano Banana Pro (Gemini 3 Pro Image) | 62.1 | 1480 ±4 | #12 | $4.50 | |
| #92 | Nano Banana Pro (Gemini 3 Pro Image Preview) | 62.1 | 1480 ±4 | #13 | $4.50 | |
| #93 | o3 Mini | OpenAI | 61.9 | 1319 ±4 | #103 | $1.93 |
| #94 | Claude Sonnet 5 | Anthropic | 60.5 | 1442 ±5 | #35 | $4.00 |
| #95 | GPT-5 | OpenAI | 60 | 1406 ±4 | #66 | $3.44 |
| #96 | GPT-5 Chat | OpenAI | 59.3 | 1404 ±4 | #68 | $3.44 |
| #97 | GPT-5.6 Terra | OpenAI | 54.7 | 1446 ±5 | #32 | $4.50 |
| #98 | GPT-5.2 Chat | OpenAI | 49.6 | 1439 ±4 | #39 | $4.81 |
| #99 | Gemma 2 27B | 48.2 | 1231 ±3 | #121 | $0.65 | |
| #100 | GPT-5.4 | OpenAI | 44.9 | 1453 ±4 | #25 | $5.63 |
| #101 | GPT-5.2 | OpenAI | 44.1 | 1412 ±3 | #63 | $4.81 |
| #102 | Claude Sonnet 4.6 | Anthropic | 43 | 1458 ±4 | #19 | $6.00 |
| #103 | Claude Sonnet 4.5 | Anthropic | 39.7 | 1438 ±3 | #41 | $6.00 |
| #104 | GPT-5.3 Chat | OpenAI | 39.1 | 1388 ±4 | #75 | $4.81 |
| #105 | Claude Opus 5 | Anthropic | 30.5 | 1505 ±4 | #2 | $10.00 |
| #106 | Command A | Cohere | 29.9 | 1331 ±3 | #100 | $4.38 |
| #107 | Claude Opus 4.6 | Anthropic | 29.8 | 1498 ±3 | #3 | $10.00 |
| #108 | Claude Opus 4.7 | Anthropic | 28.3 | 1483 ±4 | #7 | $10.00 |
| #109 | Claude Opus 4.8 | Anthropic | 25.3 | 1453 ±4 | #24 | $10.00 |
| #110 | Claude Opus 4.5 | Anthropic | 25.1 | 1451 ±3 | #28 | $10.00 |
| #111 | GPT-5.5 | OpenAI | 23.6 | 1466 ±4 | #18 | $11.25 |
| #112 | Claude Sonnet 4 | Anthropic | 23.2 | 1339 ±4 | #97 | $6.00 |
| #113 | Mistral Large 2407open | Mistral | 22 | 1266 ±4 | #117 | $3.00 |
| #114 | GPT-4o (2024-08-06) | OpenAI | 18.8 | 1282 ±4 | #112 | $4.38 |
| #115 | GPT-4.1 | OpenAI | 18.1 | 1263 ±4 | #118 | $3.50 |
| #116 | Claude Fable 5.1 | Anthropic | 15.4 | 1508 ±8 | #1 | $20.00 |
| #117 | Claude Fable 5 | Anthropic | 14.6 | 1493 ±5 | #5 | $20.00 |
| #118 | GPT-4o (2024-05-13) | OpenAI | 13.4 | 1300 ±3 | #106 | $7.50 |
| #119 | GPT-6 Astra | OpenAI | 12.2 | 1444 ±12 | #34 | $20.00 |
| #120 | Claude Opus 4.1 | Anthropic | 7.3 | 1418 ±3 | #55 | $30.00 |
| #121 | Command R+ (08-2024) | Cohere | 6.6 | 1229 ±7 | #122 | $4.38 |
| #122 | Claude Opus 4 | Anthropic | 5.5 | 1366 ±4 | #87 | $30.00 |
| #123 | GPT-4 Turbo | OpenAI | 4.8 | 1272 ±4 | #116 | $15.00 |
Methodology
Value Score = (Arena Score − 1200) ÷ blended price per 1M tokens, where blended price = 0.75 × input + 0.25 × output (a typical 3:1 token mix). The 1200-point baseline measures capability above a dated-model floor, so cheap-but-weak models can't win on price alone. A per-dollar metric still inherently favors efficient models over absolute frontier models — that's the point of this page; for raw capability, sort by Arena rank.
Capability data: Arena Scores © LMArena, licensed CC-BY-4.0, as of 13 Sept 2026. Prices: live market data, refreshed every 6 hours. No affiliate links on this page.