Together.ai: Models Intelligence, Performance & Price
Analysis of Together.ai's models across key metrics including quality, price, output speed, latency, context window & more. This analysis is intended to support you in choosing the best model provided by Together.ai for your use-case. For more details including relating to our methodology, see our FAQs. Models analyzed: Llama 3.3 70B Turbo, Llama 3.1 405B Turbo, Llama 3.2 90B (Vision) Turbo, Llama 3.1 70B Turbo, Llama 3.2 11B (Vision) Turbo, Llama 3.1 8B Turbo, Llama 3.2 3B Turbo, Gemma 2 27B, Gemma 2 9B, Mistral Small 3, Mixtral 8x22B, Mixtral 8x7B, DeepSeek R1, DeepSeek R1 Distill Llama 70B, DeepSeek V3 (Mar' 25), DeepSeek R1 Distill Qwen 14B, DeepSeek R1 Distill Qwen 1.5B, Qwen2.5 72B Turbo, Qwen2.5 Coder 32B, Qwen2 72B, QwQ-32B, Llama 3 70B (Reference, FP16), Llama 3 70B (Turbo, FP8), Llama 3 8B, Mistral 7B, DeepSeek V3 (Dec '24) (FP8), and QwQ 32B-Preview.
Link:
Together.ai Model Comparison Summary
Intelligence:
DeepSeek R1 and 
QwQ-32B are the highest quality models offered by Together.ai, followed by DeepSeek V3 (Mar' 25), DeepSeek R1 Distill Qwen 14B & DeepSeek R1 Distill Llama 70B.Output Speed (tokens/s): DeepSeek R1 Distill Qwen 1.5B (385 t/s) and 
Llama 3.1 8B Turbo (314 t/s) are the fastest models offered by Together.ai, followed by Llama 3 8B, 
Mistral 7B & DeepSeek R1 Distill Qwen 14B.Latency (seconds): Mistral 7B (0.17s) and DeepSeek R1 Distill Qwen 1.5B (0.20s) are the lowest latency models offered by Together.ai, followed by Llama 3.1 8B Turbo, Mistral Small 3 & 
Gemma 2 9B.Blended Price ($/M tokens): Llama 3.2 3B Turbo ($0.06) and Llama 3.2 11B (Vision) Turbo ($0.18) are the cheapest models offered by Together.ai, followed by Llama 3.1 8B Turbo, DeepSeek R1 Distill Qwen 1.5B & Llama 3 8B.Context Window Size: Qwen2.5 72B Turbo (131k) and Qwen2.5 Coder 32B (131k) are the largest context window models offered by Together.ai, followed by QwQ-32B, Llama 3.3 70B Turbo & Llama 3.1 405B Turbo.
DeepSeek R1 and QwQ-32B are the highest quality models offered by Together.ai, followed by DeepSeek V3 (Mar' 25), DeepSeek R1 Distill Qwen 14B & DeepSeek R1 Distill Llama 70B.Output Speed (tokens/s): DeepSeek R1 Distill Qwen 1.5B (385 t/s) and Llama 3.1 8B Turbo (314 t/s) are the fastest models offered by Together.ai, followed by Llama 3 8B, Mistral 7B & DeepSeek R1 Distill Qwen 14B.Latency (seconds): Mistral 7B (0.17s) and DeepSeek R1 Distill Qwen 1.5B (0.20s) are the lowest latency models offered by Together.ai, followed by Llama 3.1 8B Turbo, Mistral Small 3 & Gemma 2 9B.Blended Price ($/M tokens): Llama 3.2 3B Turbo ($0.06) and Llama 3.2 11B (Vision) Turbo ($0.18) are the cheapest models offered by Together.ai, followed by Llama 3.1 8B Turbo, DeepSeek R1 Distill Qwen 1.5B & Llama 3 8B.Context Window Size: Qwen2.5 72B Turbo (131k) and Qwen2.5 Coder 32B (131k) are the largest context window models offered by Together.ai, followed by QwQ-32B, Llama 3.3 70B Turbo & Llama 3.1 405B Turbo.Highlights
Intelligence
Artificial Analysis Intelligence Index; Higher is better
Speed
Output Tokens per Second; Higher is better
Price
USD per 1M Tokens; Lower is better
Parallel Queries:
Prompt Length:
Features | Model Intelligence | Price | Output tokens/s | Latency | End-to-End Response Time | ||||
|---|---|---|---|---|---|---|---|---|---|
Further Analysis | |||||||||
 | DeepSeek R1 | 128k | 60 | $4.00 | 83.6 | 0.54 | 34.59 | 28.08 | |
| QwQ-32B | 131k | 58 | $1.20 | 97.8 | 0.47 | 31.05 | 25.47 | |
| DeepSeek V3 (Mar' 25) | 128k | 53 | $1.25 | 35.4 | 0.63 | 14.74 | N/A | |
| DeepSeek R1 Distill Qwen 14B | 128k | 49 | $1.60 | 167.0 | 0.25 | 15.22 | 11.98 | |
| DeepSeek R1 Distill Llama 70B | 128k | 48 | $2.00 | 111.7 | 0.29 | 22.67 | 17.90 | |
| DeepSeek V3 (Dec '24) (FP8) | 128k | 46 | $1.25 | 33.0 | 0.64 | 15.80 | N/A | |
| QwQ 32B-Preview | 33k | 43 | $1.20 | 97.2 | 0.42 | 26.15 | 20.58 | |
| Llama 3.3 70B Turbo | 128k | 41 | $0.88 | 136.3 | 0.35 | 4.02 | N/A | |
| Llama 3.1 405B Turbo | 128k | 40 | $3.50 | 106.3 | 0.51 | 5.21 | N/A | |
| Qwen2.5 72B Turbo | 131k | 40 | $1.20 | 89.3 | 0.49 | 6.09 | N/A | |
| Qwen2.5 Coder 32B | 131k | 36 | $0.80 | 73.4 | 0.53 | 7.35 | N/A | |
| Llama 3.1 70B Turbo | 128k | 35 | $0.88 | 129.8 | 0.46 | 4.31 | N/A | |
| Mistral Small 3 | 32k | 35 | $0.80 | 96.9 | 0.22 | 5.38 | N/A | |
| Llama 3.2 90B (Vision) Turbo | 128k | 33 | $1.20 | 27.9 | 0.25 | 18.14 | N/A | |
| Qwen2 72B | 33k | 33 | $0.90 | 65.9 | 0.39 | 7.97 | N/A | |
| Llama 3 70B (Reference, FP16) | 8k | 27 | $0.90 | 154.7 | 0.48 | 3.71 | N/A | |
| Llama 3 70B (Turbo, FP8) | 8k | 27 | $0.88 | 21.5 | 0.25 | 23.54 | N/A | |
| Mixtral 8x22B | 65k | 26 | $1.20 | 71.2 | 0.37 | 7.39 | N/A | |
| Llama 3.1 8B Turbo | 128k | 24 | $0.18 | 314.2 | 0.20 | 1.79 | N/A | |
| Gemma 2 9B | 8k | 22 | $0.30 | 134.3 | 0.22 | 3.95 | N/A | |
| Llama 3 8B | 8k | 21 | $0.20 | 246.9 | 0.31 | 2.33 | N/A | |
| Llama 3.2 3B Turbo | 128k | 20 | $0.06 | 158.7 | 0.30 | 3.45 | N/A | |
| DeepSeek R1 Distill Qwen 1.5B | 128k | 19 | $0.18 | 384.8 | 0.20 | 6.69 | 5.20 | |
| Mixtral 8x7B | 33k | 17 | $0.60 | 53.2 | 0.33 | 9.73 | N/A | |
| Mistral 7B | 8k | 10 | $0.20 | 173.4 | 0.17 | 3.05 | N/A | |
| Llama 3.2 11B (Vision) Turbo | 128k | $0.18 | 115.6 | 0.23 | 4.55 | N/A | ||
| Gemma 2 27B | 8k | $0.80 | 90.4 | 0.24 | 5.77 | N/A | ||
Key definitions
Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).
Latency (Time to First Token): Time to first token received, in seconds, after API request sent. For reasoning models which share reasoning tokens, this will be the first reasoning token. For models which do not support streaming, this represents time to receive the completion.
Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Output Price: Price per token generated by the model (received from the API), represented as USD per million Tokens.
Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Time period: Metrics are 'live' and are based on the past 72 hours of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.