PRICE THE WORK, NOT THE HYPE.
Know what it costs.
One workload, applied consistently across documented API prices.
- No alternative here with both equal-or-higher task fit and equal-or-lower cost
- Other scored models with an eligible price
9 choices sit on that frontier in this plot. Unscored models are excluded from the plot, not from the catalogue or the price table.
| Model | Per workload | Per month | Basis | Select |
|---|---|---|---|---|
| Dots3-Note PreviewDots Studio | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| GLM 4.7 FlashZ.ai | $0.00 | $0.00 | verifiedSep 8, 2026 | |
| LFM2.5-2.6BLiquidAI | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| Ling 3.0 Flash SanteinclusionAI | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| Nemotron 3 Nano OmniNVIDIA | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| Nex-N2.5-MiniNex AGI | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| Nex-N2.5-ProNex AGI | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| North Mini CodeCohere | $0.00 | $0.00 | verifiedSep 11, 2026 | |
| Ling 3.0 FlashinclusionAI | $0.02 | $2.18 | verifiedSep 11, 2026 | |
| Mistral NemoMistral | $0.03 | $2.50 | verifiedSep 11, 2026 | |
| Granite 4.0 MicroIBM | $0.04 | $3.94 | verifiedSep 11, 2026 | |
| Mercury 2.5Inception | $0.04 | $4.48 | verifiedSep 11, 2026 | |
| Llama 3.1 8B InstructMeta | $0.05 | $4.85 | verifiedSep 11, 2026 | |
| gpt-oss-20bOpenAI | $0.06 | $5.60 | verifiedSep 11, 2026 | |
| Qwen3.7 FlashQwen | $0.06 | $5.60 | verifiedSep 8, 2026 | |
| Ministral 3 3B 2512Mistral | $0.06 | $5.70 | verifiedSep 8, 2026 | |
| Ling 3.0 Flash FininclusionAI | $0.06 | $6.24 | verifiedSep 11, 2026 | |
| Ling 3.0 Flash VLinclusionAI | $0.06 | $6.24 | verifiedSep 11, 2026 | |
| Laguna XS 2.1Poolside | $0.06 | $6.30 | verifiedSep 11, 2026 | |
| Nova Micro 1.0Amazon | $0.06 | $6.30 | verifiedSep 11, 2026 |
207 models excluded from numerical estimates
Missing, stale, unsupported, or incompatible pricing is not treated as free.
Qwen3.8-Flash-Next — Price not verified. Choose “Include saved prices” to use older catalogue rates.
ByteDance Seed 2.1 Pro Preview — Price not verified. Choose “Include saved prices” to use older catalogue rates.
Claude 3 Haiku — Claude API retired 2026-04-20. This does not establish retirement or pricing on every partner endpoint.
Claude Opus 4 — Claude API retired 2026-06-15. Pricing documentation still lists Google Cloud availability; partner route status and prices are not verified here.
Claude Opus 4.1 — Claude API retired 2026-08-05. Pricing documentation still lists availability on Bedrock and Google Cloud; partner route status and prices are not verified here.
Claude Sonnet 4 — Claude API retired 2026-06-15. Pricing documentation still lists Bedrock and Google Cloud availability; partner route status and prices are not verified here.
R1 Distill Llama 70B — The per-request prompt exceeds this model’s documented context window.
Nano Banana Pro (Gemini 3 Pro Image) — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
GPT-5.4 Image 2 — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
GPT-5.6 Luna Pro — This route has conditional or scheduled pricing. See the source for the applicable rate.
GPT-5.6 Sol Pro — This route has conditional or scheduled pricing. See the source for the applicable rate.
GPT-5.6 Terra Pro — This route has conditional or scheduled pricing. See the source for the applicable rate.
GPT-6 Astra Pro — This route has conditional or scheduled pricing. See the source for the applicable rate.
Fugu Ultra v2 — This route has conditional or scheduled pricing. See the source for the applicable rate.
FLUX Video Edit — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
DeepSeek V4.1 Flash — This route has conditional or scheduled pricing. See the source for the applicable rate.
GPT Image 2.5 Sunburst — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
GPT Image 2.5 Flare — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
MAI-Image-2.6 — The per-request prompt exceeds this model’s documented context window.
MAI-Image-2.6 Flash — The per-request prompt exceeds this model’s documented context window.
MAI-Transcribe 2 — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
H3 Max — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Wan 3.0 Prime — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Muse Image — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Recraft V4 Styles Pro — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Recraft V4 Styles Vector — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Recraft V4 Styles Pro Vector — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Recraft V4 Styles — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Wan 3.0 — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.
Avatar IV — Use this model’s linked rate card. Text-token workload estimates do not cover media, transcription, embeddings or reranking.