The Known Good Updated 25 Jul 2026
Providers· Inference host· 18 tracked models

Together — pricing & performance

Compare models All providers Data sources
Models served
18
of 908 tracked models
Mean output speed
63
tokens per second, averaged over
17 measured endpoints
Mean latency
0.6s
time to first token, averaged over
17 measured endpoints
Cheapest blended
$0.075
per 1M tokens at a 3:1
input:output blend
Output speed on Together

Median output tokens per second for each model this host serves, derived from OpenRouter endpoint telemetry. Higher is better. We do not run these measurements ourselves.

12 of 18 models ▾ ☰⚙⇩ +Add model from specific provider
130
111
97
95
88
86
67
62
59
53
50
46
The Known Good
18 models on Together
Full leaderboard →
Model Creator Input $/M Output $/M Blended Speed TTFT Context
Arcee AI: Virtuoso Large arcee-ai $0.75 $1.20 $0.86 — — 131k
Deep Cogito: Cogito v2.1 671B deepcogito $1.25 $1.25 $1.25 25 0.3s 128k
DeepSeek: DeepSeek V4 Pro DeepSeek $1.74 $3.48 $2.17 53 0.8s 512k
Google: Gemma 3n 4B Google $0.060 $0.12 $0.075 46 0.8s 33k
Google: Gemma 4 31B Google $0.28 $0.86 $0.42 37 0.6s 262k
Meta: Llama 3.3 70B Instruct Meta $1.04 $1.04 $1.04 18 1.2s 131k
Meta: Llama Guard 4 12B Meta $0.20 $0.20 $0.20 13 0.2s 1.05M
MiniMax: MiniMax M2.7 MiniMax $0.30 $1.20 $0.53 97 0.4s 197k
MiniMax: MiniMax M3 MiniMax $0.30 $1.20 $0.53 50 0.6s 524k
MoonshotAI: Kimi K2.6 Moonshot AI $1.20 $4.50 $2.02 130 0.4s 262k
MoonshotAI: Kimi K2.7 Code Moonshot AI $0.95 $4.00 $1.71 67 0.6s 262k
NVIDIA: Nemotron 3 Ultra NVIDIA $0.60 $3.60 $1.35 88 1.2s 512k
OpenAI: gpt-oss-120b OpenAI $0.15 $0.60 $0.26 111 0.2s 131k
OpenAI: gpt-oss-20b OpenAI $0.050 $0.20 $0.087 95 0.3s 131k
Qwen: Qwen2.5 7B Instruct Alibaba $0.30 $0.30 $0.30 59 0.3s 33k
Qwen: Qwen3.5-9B Alibaba $0.17 $0.25 $0.19 33 1.0s 262k
Thinking Machines: Inkling Thinking Machines $1.00 $4.05 $1.76 86 0.4s 524k
Z.ai: GLM 5.2 Z.ai $1.40 $4.40 $2.15 62 0.9s 262k
The Known Good
Source. Prices, endpoint availability and throughput for Together are ingested from OpenRouter; model metadata is enriched from models.dev. A blank cell means the figure was not published for that endpoint — it never means zero. Blended price is (3 × input + output) ÷ 4, the same 3:1 blend used everywhere on this site; see methodology. Quality scores are not provider-specific: a model scores the same wherever it is hosted, so quality lives on the model pages.