AI model value ranking (value_score)
Source: official OpenRouter model registry (recomputed daily, 2026-11-09). Every figure comes from our structured dataset; the formula ships per row.
Full-mark models
15
Context >= 1M
196
Median paid cost
$0.88/M
Cheapest paid
openai/gpt-oss-120b $0.0173/M
| Model | Provider | Blended $/M | Context | Cost eff | Ctx eff | value_score |
|---|---|---|---|---|---|---|
| ~z-ai/glm-flash-latest | ~z-ai | 0.0445 | 1.0M | 1.00 | 1.00 | 100 |
| z-ai/glm-5.3-flash:batch | z-ai | 0.0825 | 1.0M | 1.00 | 1.00 | 100 |
| openai/gpt-6-luna-pro:batch | openai | 0.1000 | 1.1M | 1.00 | 1.00 | 100 |
| openai/gpt-6-luna:batch | openai | 0.1000 | 1.1M | 1.00 | 1.00 | 100 |
| deepseek/deepseek-v4-flash | deepseek | 0.1138 | 1.0M | 1.00 | 1.00 | 100 |
| deepseek/deepseek-v4-flash-0731 | deepseek | 0.1192 | 1.0M | 1.00 | 1.00 | 100 |
| ~deepseek/deepseek-v4-flash-latest | ~deepseek | 0.1192 | 1.0M | 1.00 | 1.00 | 100 |
| meta/muse-spark-1.2-contributor | meta | 0.1250 | 1.0M | 1.00 | 1.00 | 100 |
| meta/muse-spark-1.3-contributor | meta | 0.1250 | 1.0M | 1.00 | 1.00 | 100 |
| ~deepseek/deepseek-flash-latest | ~deepseek | 0.1295 | 1.0M | 1.00 | 1.00 | 100 |
| meta-llama/llama-4-scout | meta-llama | 0.1500 | 1.3M | 1.00 | 1.00 | 100 |
| google/lyria-3-clip-preview | 0.0000 | 1.0M | 1.00 | 1.00 | 100 |
Three key readings
- Cheap long-context is a red ocean: 15 models score full marks (cost <= $0.15/M and context near 1M), led by ~z-ai/glm-flash-latest. The constraint has shifted from "can't afford it" to "which one is reliable".
- Batch endpoints are the hidden discount: full-mark rows ending in :batch are batch-priced; the same model's instant endpoint often costs several times more - hourly-latency workloads should check batch first.
- The median is your budget baseline: median paid blended cost is $0.88/M, while 16 free/zero-priced models sit outside that median - don't budget off the full-mark leaderboard.
Use this data
Machine-readable endpoints: GET /api/supply/ai-model-value/digest (free one-liner),
GET /api/supply/ai-model-value/smart_top20 ($0.05/call),
GET /api/supply/ai-model-value/best_under_budget?budget=0.05&min_context=128000 ($0.05/call).
Source: official OpenRouter model registry; the value_score formula and components ship per row (recomputed daily, 2026-11-09)