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 |
|---|---|---|---|---|---|---|
| 🥇 | gpt-oss-120b | OpenAI | 3010 | 1381 ±8 | #67 | $0.06 |
| 🥈 | Qwen3 30B A3B Instruct 2507open | Qwen | 2575.1 | 1417 ±9 | #51 | $0.08 |
| 🥉 | DeepSeek V4 Flashopen | DeepSeek | 2243.6 | 1452 ±7 | #32 | $0.11 |
| 4 | Gemma 4 26B A4B | 1913.7 | 1444 ±15 | #37 | $0.13 | |
| 5 | gpt-oss-20b | OpenAI | 1913.7 | 1309 ±13 | #86 | $0.06 |
| 6 | Qwen3.5-Flashopen | Qwen | 1865.5 | 1412 ±6 | #54 | $0.11 |
| 7 | Gemma 4 31B | 1431.5 | 1454 ±15 | #29 | $0.18 | |
| 8 | Gemma 3 27B | 1225 | 1323 ±7 | #84 | $0.1 | |
| 9 | Qwen3 32Bopen | Qwen | 1216.9 | 1358 ±24 | #79 | $0.13 |
| 10 | GPT-5 Nano | OpenAI | 1102.5 | 1352 ±15 | #80 | $0.14 |
| 11 | Gemma 3 12B | 1089.3 | 1282 ±23 | #93 | $0.08 | |
| 12 | Gemini 2.5 Flash Lite | 992.6 | 1374 ±8 | #72 | $0.18 | |
| 13 | Gemini 2.5 Flash Lite Preview 09-2025 | 980 | 1372 ±6 | #74 | $0.18 | |
| 14 | DeepSeek V3.2open | DeepSeek | 965.4 | 1449 ±6 | #35 | $0.26 |
| 15 | Gemma 3n 4B | 929.3 | 1270 ±10 | #95 | $0.08 | |
| 16 | MiniMax M2.5open | MiniMax | 851.9 | 1379 ±7 | #69 | $0.21 |
| 17 | Mistral Small 3open | Mistral | 803.5 | 1246 ±12 | #100 | $0.06 |
| 18 | MiniMax M2.7open | MiniMax | 801.9 | 1453 ±7 | #31 | $0.32 |
| 19 | Qwen3 Next 80B A3B Thinkingopen | Qwen | 712.7 | 1391 ±11 | #64 | $0.27 |
| 20 | Qwen3 Next 80B A3B Instructopen | Qwen | 704.2 | 1441 ±9 | #38 | $0.34 |
| 21 | Qwen3 VL 235B A22B Instructopen | Qwen | 649.2 | 1440 ±13 | #39 | $0.37 |
| 22 | Qwen3 30B A3Bopen | Qwen | 641.4 | 1338 ±9 | #81 | $0.22 |
| 23 | GPT-4.1 Nano | OpenAI | 606.9 | 1306 ±19 | #87 | $0.18 |
| 24 | Qwen3.5-35B-A3Bopen | Qwen | 588.7 | 1409 ±7 | #55 | $0.36 |
| 25 | Llama 4 Scoutopen | Meta | 575.3 | 1286 ±9 | #91 | $0.15 |
| 26 | MiniMax M3open | MiniMax | 523.8 | 1475 ±9 | #17 | $0.52 |
| 27 | Qwen3.7 Plusopen | Qwen | 509.6 | 1485 ±10 | #10 | $0.56 |
| 28 | DeepSeek V4 Proopen | DeepSeek | 495.4 | 1469 ±7 | #20 | $0.54 |
| 29 | Gemma 3 4B | 492.8 | 1231 ±23 | #102 | $0.06 | |
| 30 | DeepSeek V3.1 Terminusopen | DeepSeek | 469.5 | 1407 ±21 | #58 | $0.44 |
| 31 | Qwen3 235B A22B Thinking 2507open | Qwen | 460.2 | 1424 ±15 | #48 | $0.49 |
| 32 | Llama 3.3 70B Instructopen | Meta | 445.2 | 1269 ±7 | #96 | $0.16 |
| 33 | GPT-5.4 Nano | OpenAI | 436.5 | 1402 ±7 | #59 | $0.46 |
| 34 | MiniMax M2.1open | MiniMax | 423.8 | 1423 ±10 | #50 | $0.52 |
| 35 | Qwen3.5-27Bopen | Qwen | 420.1 | 1425 ±7 | #47 | $0.54 |
| 36 | Llama 4 Maverickopen | Meta | 392.8 | 1303 ±7 | #88 | $0.26 |
| 37 | MiniMax M2open | MiniMax | 390.1 | 1374 ±15 | #71 | $0.45 |
| 38 | Qwen3.6 Plusopen | Qwen | 363.4 | 1466 ±7 | #23 | $0.73 |
| 39 | Phi 4 | Microsoft | 363.4 | 1232 ±10 | #101 | $0.09 |
| 40 | Gemini 3.1 Flash Lite | 355.2 | 1400 ±6 | #62 | $0.56 | |
| 41 | Kimi K2.5open | Moonshot AI | 346.9 | 1473 ±6 | #18 | $0.79 |
| 42 | Qwen3 Coder 480B A35Bopen | Qwen | 345.5 | 1413 ±9 | #53 | $0.62 |
| 43 | GPT-4o-mini (2024-07-18) | OpenAI | 343.6 | 1290 ±7 | #90 | $0.26 |
| 44 | Qwen3.5-122B-A10Bopen | Qwen | 330.8 | 1437 ±7 | #41 | $0.72 |
| 45 | Qwen-Plusopen | Qwen | 329.7 | 1329 ±18 | #83 | $0.39 |
| 46 | GPT-5 Mini | OpenAI | 301.1 | 1407 ±8 | #57 | $0.69 |
| 47 | Qwen3.5 397B A17Bopen | Qwen | 295.7 | 1467 ±6 | #22 | $0.9 |
| 48 | Mistral Small 3.1 24Bopen | Mistral | 273.4 | 1310 ±8 | #85 | $0.4 |
| 49 | Qwen3 VL 235B A22B Thinkingopen | Qwen | 269.9 | 1428 ±14 | #45 | $0.85 |
| 50 | Gemini 2.5 Flash | 263.1 | 1424 ±4 | #49 | $0.85 | |
| 51 | R1 0528open | DeepSeek | 248.8 | 1427 ±11 | #46 | $0.91 |
| 52 | GPT-4.1 Mini | OpenAI | 239.3 | 1368 ±7 | #76 | $0.7 |
| 53 | Kimi K2 Thinkingopen | Moonshot AI | 235.6 | 1453 ±6 | #30 | $1.08 |
| 54 | Gemini 3 Flash Preview | 233.6 | 1463 ±8 | #24 | $1.13 | |
| 55 | Qwen3 235B A22Bopen | Qwen | 233.3 | 1386 ±9 | #66 | $0.8 |
| 56 | Nova 2 Lite | Amazon | 221.2 | 1388 ±12 | #65 | $0.85 |
| 57 | Kimi K2.6open | Moonshot AI | 210.9 | 1484 ±7 | #12 | $1.35 |
| 58 | MiniMax M1open | MiniMax | 187.1 | 1359 ±8 | #78 | $0.85 |
| 59 | Kimi K2 0905open | Moonshot AI | 185.8 | 1400 ±12 | #63 | $1.08 |
| 60 | Kimi K2 0711open | Moonshot AI | 177.3 | 1378 ±8 | #70 | $1.00 |
| 61 | Grok 4.20 | xAI | 167.6 | 1462 ±6 | #25 | $1.56 |
| 62 | Grok 4.20 Multi-Agent | xAI | 164.9 | 1458 ±6 | #27 | $1.56 |
| 63 | Qwen3 Maxopen | Qwen | 164.7 | 1457 ±8 | #28 | $1.56 |
| 64 | Qwen3.7 Maxopen | Qwen | 159.4 | 1499 ±18 | #5 | $1.88 |
| 65 | Llama 3.1 70B Instructopen | Meta | 150.5 | 1260 ±7 | #98 | $0.4 |
| 66 | R1open | DeepSeek | 149.5 | 1372 ±12 | #73 | $1.15 |
| 67 | GPT-5.4 Mini | OpenAI | 141.8 | 1439 ±6 | #40 | $1.69 |
| 68 | Grok 4.3 | xAI | 138.7 | 1417 ±7 | #52 | $1.56 |
| 69 | Claude Haiku 4.5 | Anthropic | 125.1 | 1450 ±5 | #34 | $2.00 |
| 70 | Qwen3.6 Max Previewopen | Qwen | 114.1 | 1467 ±15 | #21 | $2.34 |
| 71 | o4 Mini | OpenAI | 87.6 | 1369 ±7 | #75 | $1.93 |
| 72 | Gemini 3.5 Flash | 86.9 | 1493 ±10 | #8 | $3.38 | |
| 73 | Mistral Medium 3.5open | Mistral | 85.9 | 1458 ±11 | #26 | $3.00 |
| 74 | o3 Mini | OpenAI | 84.3 | 1362 ±6 | #77 | $1.93 |
| 75 | Gemini 2.5 Pro | 73.4 | 1452 ±4 | #33 | $3.44 | |
| 76 | Claude Sonnet 5 | Anthropic | 69.6 | 1478 ±18 | #16 | $4.00 |
| 77 | GPT-5.1 | OpenAI | 68.7 | 1436 ±7 | #42 | $3.44 |
| 78 | GPT-5 | OpenAI | 68.3 | 1435 ±8 | #43 | $3.44 |
| 79 | Gemini 3.1 Pro Preview Custom Tools | 63.3 | 1485 ±6 | #11 | $4.50 | |
| 80 | Nano Banana Pro (Gemini 3 Pro Image) | 62.8 | 1483 ±7 | #13 | $4.50 | |
| 81 | GPT-5 Chat | OpenAI | 58.2 | 1400 ±8 | #61 | $3.44 |
| 82 | GPT-5.2 Chat | OpenAI | 51.3 | 1447 ±7 | #36 | $4.81 |
| 83 | Claude Sonnet 4.6 | Anthropic | 50.3 | 1502 ±6 | #4 | $6.00 |
| 84 | GPT-5.4 | OpenAI | 49.8 | 1480 ±6 | #14 | $5.63 |
| 85 | GPT-5.2 | OpenAI | 48.2 | 1432 ±6 | #44 | $4.81 |
| 86 | Claude Sonnet 4.5 | Anthropic | 47.6 | 1485 ±5 | #9 | $6.00 |
| 87 | GPT-5.3 Chat | OpenAI | 43 | 1407 ±7 | #56 | $4.81 |
| 88 | Claude Opus 4.6 | Anthropic | 33.5 | 1535 ±6 | #1 | $10.00 |
| 89 | Claude Opus 4.7 | Anthropic | 31.8 | 1518 ±7 | #3 | $10.00 |
| 90 | Claude Sonnet 4 | Anthropic | 30 | 1380 ±7 | #68 | $6.00 |
| 91 | Command A | Cohere | 29.9 | 1331 ±6 | #82 | $4.38 |
| 92 | Claude Opus 4.5 | Anthropic | 29.8 | 1498 ±6 | #6 | $10.00 |
| 93 | Claude Opus 4.8 | Anthropic | 29.4 | 1494 ±9 | #7 | $10.00 |
| 94 | Mistral Large 2407open | Mistral | 25.7 | 1277 ±8 | #94 | $3.00 |
| 95 | GPT-5.5 | OpenAI | 24.8 | 1479 ±7 | #15 | $11.25 |
| 96 | GPT-4o (2024-08-06) | OpenAI | 19.2 | 1284 ±8 | #92 | $4.38 |
| 97 | Gemma 2 27B | 17.7 | 1212 ±6 | #103 | $0.65 | |
| 98 | Claude Fable 5 | Anthropic | 16.5 | 1530 ±18 | #2 | $20.00 |
| 99 | GPT-4.1 | OpenAI | 15.6 | 1255 ±7 | #99 | $3.50 |
| 100 | GPT-4o (2024-05-13) | OpenAI | 13 | 1298 ±6 | #89 | $7.50 |
| 101 | Claude Opus 4.1 | Anthropic | 9.1 | 1473 ±5 | #19 | $30.00 |
| 102 | Claude Opus 4 | Anthropic | 6.7 | 1402 ±7 | #60 | $30.00 |
| 103 | GPT-4 Turbo | OpenAI | 4.6 | 1269 ±7 | #97 | $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 2 Jul 2026. Prices: live market data, refreshed every 6 hours. No affiliate links on this page.