PRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.012 to 0.00779mPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.012 to 0.00779mPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.012 to 0.00779mPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.012 to 0.00779mPRICEdeepseek/deepseek-v3.1-terminus input_per_mtok decreased from 0.30 to 0.279mPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok increased from 0.0011 to 0.01240mPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok increased from 0.60 to 1.2840mPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.013 to 0.01240mPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.017 to 0.01240mPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.017 to 0.01240mNEW MODELnvidia/switchyard appeared in the catalogue40mPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok decreased from 1.04 to 0.601hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok increased from 0.0038 to 0.0131hPRICEnvidia/nemotron-3-ultra-550b-a55b cached_input_per_mtok decreased from 0.12 to 0.101hPRICEnvidia/nemotron-3-ultra-550b-a55b output_per_mtok decreased from 2.40 to 2.201hPRICEnvidia/nemotron-3-ultra-550b-a55b input_per_mtok decreased from 0.60 to 0.501hPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok increased from 0.0051 to 0.0171hPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok increased from 0.0051 to 0.0171hPRICE~moonshotai/kimi-latest cached_input_per_mtok decreased from 0.80 to 0.292hPRICE~moonshotai/kimi-latest output_per_mtok decreased from 13.00 to 11.362hPRICE~moonshotai/kimi-latest input_per_mtok decreased from 1.39 to 1.002hPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok increased from 0.95 to 1.042hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.0058 to 0.00382hPRICEinclusionai/ling-3.0-flash-fin cached_input_per_mtok decreased from 0.012 to 0.00842hPRICEinclusionai/ling-3.0-flash-fin output_per_mtok decreased from 0.18 to 0.122hPRICEinclusionai/ling-3.0-flash-fin input_per_mtok decreased from 0.06 to 0.0422hPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.0077 to 0.00512hPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.0077 to 0.00512hPRICE~z-ai/glm-flash-latest output_per_mtok increased from 0.60 to 0.932hPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok decreased from 1.20 to 0.952hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.0086 to 0.00582hPRICEz-ai/glm-5.2 cached_input_per_mtok decreased from 0.32 to 0.262hPRICEdeepseek/deepseek-v4-flash-0731 cached_input_per_mtok decreased from 0.012 to 0.00772hPRICEdeepseek/deepseek-v4-flash-0731 input_per_mtok decreased from 0.012 to 0.00772hPRICE~moonshotai/kimi-latest input_per_mtok increased from 1.34 to 1.393hPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.0038 to 0.00113hPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok increased from 0.0038 to 0.00863hPRICEgoogle/gemma-4-26b-a4b-it cached_input_per_mtok decreased from 0.05 to 0.0383hPRICEgoogle/gemma-4-26b-a4b-it output_per_mtok decreased from 0.30 to 0.233hPRICEgoogle/gemma-4-26b-a4b-it input_per_mtok decreased from 0.09 to 0.0683h

Together AI — AI inference benchmarks

US · serverless · 6 benchmarked models · 1 region
#7 GLOBALAdd to compare
Overall85.8Rel.98.6%Tools99.7%JSON96.8%TTFT572 msP951625 msTok/s130Price$0.67 / $1.12Δ30d↑ 5.5Excluded this window: 1 throttled · 0 client faults● last run 3h ago · 24 runs in 24h

Head-to-head: vs Cerebras · vs Groq · vs OpenRouter · vs Cortecs · vs Scaleway

Provider vs market

Percentile across 8 ranked providers
Reliability 98.6%P13
Tools 99.7%P25
Structured 96.8%P38
TTFT 572 msP38
Throughput 130 tok/sP75
Cost $1.12P75

Price vs performance

Together AI highlighted vs peers
879196100$4.37$8.74$13.10$17.47Output price $/M →Score ↑Cerebras — $0.75/M out · score 99.0Groq — $0.45/M out · score 98.2Mistral — $4.50/M out · score 93.2Scaleway — $1.81/M out · score 92.4Cortecs — $1.23/M out · score 92.2OpenAI — $15.60/M out · score 90.4Together — $1.12/M out · score 85.8TogetherOpenRouter — $1.20/M out · score 84.4

30-day trend

Measured daily series
Score85.8
TTFT182 ms
Tools100.0%
Reliability93.5%
Price$0.60

Key positioning

Derived from benchmark percentiles
Strengths
  • Throughput (P75)
  • Cost (P75)
Weaknesses
  • Reliability (P13)
  • Tools (P25)
  • Structured (P38)
  • TTFT (P38)
Best suited for

Together AI in the registry

  • HeadquartersUS
  • API protocolsopenai chat completions
  • Catalogued endpoints1
  • Websitewww.together.ai
  • Last verified2026-10-02

Observed models & declared pricing

Read from Together AI's own catalogue by continuous discovery — declared data, not benchmark results.

ModelContext$ in / M$ out / MObserved
agentica-org/DeepCoder-14B-Preview131k——2026-10-02
alibaba/happyhorse-1.0-i2v———2026-10-02
alibaba/happyhorse-1.0-r2v———2026-10-02
alibaba/happyhorse-1.0-t2v———2026-10-02
alibaba/happyhorse-1.1-i2v———2026-10-02
alibaba/happyhorse-1.1-r2v———2026-10-02
alibaba/happyhorse-1.1-t2v———2026-10-02
allenai/Molmo-7B-D-09244k——2026-10-02
arcee-ai/trinity-mini128k0.0450.152026-10-02
arize-ai/qwen-2-1.5b-instruct33k0.100.102026-10-02
BAAI/bge-base-en-v1.51k0.0080.0082026-10-02
black-forest-labs/FLUX-3———2026-10-02
black-forest-labs/FLUX.1-kontext-max———2026-10-02
black-forest-labs/FLUX.1-kontext-pro———2026-10-02
black-forest-labs/FLUX.1.1-pro———2026-10-02
black-forest-labs/FLUX.2-dev———2026-10-02
black-forest-labs/FLUX.2-flex———2026-10-02
black-forest-labs/FLUX.2-max———2026-10-02
black-forest-labs/FLUX.2-pro———2026-10-02
ByteDance-Seed/Seedream-3.0———2026-10-02
ByteDance-Seed/Seedream-4.0———2026-10-02
ByteDance/Seedance-1.0-lite———2026-10-02
ByteDance/Seedance-1.0-pro———2026-10-02
ByteDance/Seedance-2.0———2026-10-02
ByteDance/Seedance-2.5———2026-10-02
ByteDance/Seedream-5.0-lite———2026-10-02
canopylabs/orpheus-3b-0.1-ft—15.00—2026-10-02
cartesia/sonic—65.00—2026-10-02
cartesia/sonic-20k65.00—2026-10-02
cartesia/sonic-30k65.00—2026-10-02

All providers · How we benchmark