Welcome to the AI Weather Report for July 21, 2026. This daily report ranks the top 10 AI models for coding by bang for the buck — a combination of raw coding capability and API pricing.
📊 Today’s Top 10 Rankings
| # | Model | Provider | Capability | Cost /M tokens | Value Score |
|---|---|---|---|---|---|
| 🥇 1 | mistral-nemo | mistralai | 62/100 | $0.0272 | 2275.2 |
| 🥈 2 | ling-2.6-flash | inclusionai | 56/100 | $0.0250 | 2240.0 |
| 🥉 3 | l3-lunaris-8b | sao10k | 58/100 | $0.0475 | 1221.1 |
| 4 | mistral-small-24b-instruct-2501 | mistralai | 72/100 | $0.0725 | 993.1 |
| 5 | llama-3.1-8b-instruct | meta-llama | 62/100 | $0.0725 | 855.2 |
| 6 | mythomax-l2-13b | gryphe | 48/100 | $0.0600 | 800.0 |
| 7 | gpt-oss-20b | openai | 78/100 | $0.1050 | 742.9 |
| 8 | qwen-2.5-7b-instruct | qwen | 60/100 | $0.0850 | 705.9 |
| 9 | laguna-xs-2.1 | poolside | 72/100 | $0.1050 | 685.7 |
| 10 | gpt-oss-120b | openai | 93/100 | $0.1368 | 680.1 |
📈 Analysis
🏆 Best Value Today: mistral-nemo scores 2275.2 with a capability rating of 62 at $0.0272/M tokens.
💵 Cheapest Premium Model: ling-2.6-flash at $0.0250/M tokens (capability: 56).
What “Value Score” means: Capability score (based on SWE-bench, HumanEval, LiveCodeBench) divided by blended cost per million tokens (25% input + 75% output weights for coding workloads). Free tier models get a massive boost. Higher is better.
📋 All Scored Models (67 total)
| # | Model | Provider | Capability | Cost /M tok | Value |
|---|---|---|---|---|---|
| 1 | mistral-nemo | mistralai | 62 | $0.0272 | 2275.2 |
| 2 | ling-2.6-flash | inclusionai | 56 | $0.0250 | 2240.0 |
| 3 | l3-lunaris-8b | sao10k | 58 | $0.0475 | 1221.1 |
| 4 | mistral-small-24b-instruct-2501 | mistralai | 72 | $0.0725 | 993.1 |
| 5 | llama-3.1-8b-instruct | meta-llama | 62 | $0.0725 | 855.2 |
| 6 | mythomax-l2-13b | gryphe | 48 | $0.0600 | 800.0 |
| 7 | gpt-oss-20b | openai | 78 | $0.1050 | 742.9 |
| 8 | qwen-2.5-7b-instruct | qwen | 60 | $0.0850 | 705.9 |
| 9 | laguna-xs-2.1 | poolside | 72 | $0.1050 | 685.7 |
| 10 | gpt-oss-120b | openai | 93 | $0.1368 | 680.1 |
| 11 | gemma-3-4b-it | 50 | $0.0875 | 571.4 | |
| 12 | deepseek-v4-flash | deepseek | 91 | $0.1641 | 554.4 |
| 13 | granite-4.1-8b | ibm-granite | 48 | $0.0875 | 548.6 |
| 14 | qwen3.5-9b | qwen | 72 | $0.1375 | 523.6 |
| 15 | gemma-3-12b-it | 60 | $0.1250 | 480.0 | |
| 16 | command-r7b-12-2024 | cohere | 54 | $0.1219 | 443.1 |
| 17 | granite-4.0-h-micro | ibm-granite | 38 | $0.0882 | 430.6 |
| 18 | ministral-3b-2512 | mistralai | 42 | $0.1000 | 420.0 |
| 19 | nova-micro-v1 | amazon | 45 | $0.1137 | 395.6 |
| 20 | hy3-preview | tencent | 68 | $0.1732 | 392.5 |
| 21 | qwen3-32b | qwen | 88 | $0.2300 | 382.6 |
| 22 | qwen3-coder-30b-a3b-instruct | qwen | 84 | $0.2200 | 381.8 |
| 23 | qwen3.5-flash-02-23 | qwen | 70 | $0.2112 | 331.4 |
| 24 | qwen3-30b-a3b-instruct-2507 | qwen | 82 | $0.2500 | 328.0 |
| 25 | gpt-oss-safeguard-20b | openai | 77 | $0.2437 | 315.9 |
| 26 | mistral-small-3.2-24b-instruct | mistralai | 78 | $0.2500 | 312.0 |
| 27 | nemotron-3-nano-30b-a3b | nvidia | 50 | $0.1625 | 307.7 |
| 28 | nova-lite-v1 | amazon | 58 | $0.1950 | 297.4 |
| 29 | gemma-3-27b-it | 68 | $0.2500 | 272.0 | |
| 30 | gemma-4-26b-a4b-it | 72 | $0.2725 | 264.2 | |
| 31 | seed-1.6-flash | bytedance-seed | 64 | $0.2437 | 262.6 |
| 32 | gpt-5-nano | openai | 82 | $0.3125 | 262.4 |
| 33 | llama-3.3-70b-instruct | meta-llama | 84 | $0.3325 | 252.6 |
| 34 | gemma-4-31b-it | 74 | $0.3075 | 240.7 | |
| 35 | step-3.5-flash | stepfun | 60 | $0.2500 | 240.0 |
| 36 | nemotron-3-super-120b-a12b | nvidia | 76 | $0.3212 | 236.6 |
| 37 | laguna-m.1 | poolside | 80 | $0.3500 | 228.6 |
| 38 | seed-2.0-mini | bytedance-seed | 72 | $0.3250 | 221.5 |
| 39 | qwen3-235b-a22b-2507 | qwen | 96 | $0.4350 | 220.7 |
| 40 | llama-3.1-70b-instruct | meta-llama | 82 | $0.4000 | 205.0 |
| 41 | llama-3.2-1b-instruct | meta-llama | 30 | $0.1575 | 190.5 |
| 42 | glm-4.7-flash | z-ai | 60 | $0.3151 | 190.4 |
| 43 | gpt-4.1-nano | openai | 60 | $0.3250 | 184.6 |
| 44 | llama-3.2-3b-instruct | meta-llama | 48 | $0.2640 | 181.8 |
| 45 | ring-2.6-1t | inclusionai | 78 | $0.4875 | 160.0 |
| 46 | qwen3-next-80b-a3b-thinking | qwen | 93 | $0.6094 | 152.6 |
| 47 | gpt-4o-mini | openai | 74 | $0.4875 | 151.8 |
| 48 | ling-2.6-1t | inclusionai | 74 | $0.4875 | 151.8 |
| 49 | qwen3-next-80b-a3b-instruct | qwen | 90 | $0.6094 | 147.7 |
| 50 | deepseek-chat | deepseek | 90 | $0.6501 | 138.4 |
| 51 | command-r-08-2024 | cohere | 60 | $0.4875 | 123.1 |
| 52 | qwen3-coder | qwen | 85 | $0.8250 | 103.0 |
| 53 | qwen-2.5-coder-32b-instruct | qwen | 86 | $0.9150 | 94.0 |
| 54 | hermes-3-llama-3.1-405b | nousresearch | 78 | $1.00 | 78.0 |
| 55 | claude-3-haiku | anthropic | 72 | $1.00 | 72.0 |
| 56 | dolphin-mistral-24b-venice-edition | cognitivecomputations | 52 | $0.7250 | 71.7 |
| 57 | gpt-4.1-mini | openai | 76 | $1.30 | 58.5 |
| 58 | deepseek-r1 | deepseek | 95 | $2.05 | 46.3 |
| 59 | gemini-2.5-flash | 86 | $1.95 | 44.1 | |
| 60 | nova-pro-v1 | amazon | 70 | $2.60 | 26.9 |
| 61 | gpt-4.1 | openai | 90 | $6.50 | 13.8 |
| 62 | gpt-5 | openai | 97 | $7.81 | 12.4 |
| 63 | gemini-2.5-pro | 94 | $7.81 | 12.0 | |
| 64 | gpt-4o | openai | 88 | $8.13 | 10.8 |
| 65 | command-r-plus-08-2024 | cohere | 68 | $8.13 | 8.4 |
| 66 | claude-sonnet-4 | anthropic | 96 | $12.00 | 8.0 |
| 67 | claude-opus-4 | anthropic | 98 | $60.00 | 1.6 |
Generated 2026-07-21 02:00 UTC · Data from OpenRouter API and public benchmarks · Bang-for-Buck = Capability / Cost
