CoreWeave: Models Intelligence, Performance & Price
Analysis of CoreWeave'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 CoreWeave for your use-case.
Most Intelligent
Intelligence index
Total 24 models
Fastest
Output speed
Total 24 models
Lowest Price
Blended price (per 1M tokens)
Total 24 models
CoreWeave offers 24 models, each with different intelligence, performance, and pricing characteristics. Below is a comparison of the key metrics across models.
- For intelligence, the top models on CoreWeave are GLM-5.2 (max) (53), DeepSeek V4 Flash 0731 (max) (52), and MiniMax-M3 (45).
- For output speed, the fastest models are Nemotron 3.5 Lightning (BF16) (343 t/s), Kimi K2.7 Code (293 t/s), and Nemotron 3 Ultra (226 t/s). Speed varies significantly across models, with a 70% difference between the fastest and slowest.
- For latency, Granite 4.1 8B (0.77s), Llama 3.3 70B (0.89s), and Qwen3 30B A3B 2507 (0.95s) offer the lowest time to first answer token.
- For pricing, gpt-oss-20b (high) ($0.04), gpt-oss-20b (low) ($0.04), and gpt-oss-120b (high) ($0.04) offer the lowest blended prices per 1M tokens.
- For context window size, DeepSeek V4 Flash (high) (1M), DeepSeek V4 Flash (1M), and GLM-5.2 (max) (262k) support the largest context windows on CoreWeave.
Highlights
Intelligence Evaluations
Artificial Analysis Intelligence Index
Intelligence Evaluations
Agentic real-world work tasks, (Elo-500)/2000
Agentic tool use
Agentic coding & terminal use
Coding
Reasoning & knowledge
Scientific reasoning
Physics reasoning
Knowledge
1 - hallucination rate
Long context reasoning
Agentic knowledge work, Elo
Agentic SaaS workflows
Legal agentic work, criterion pass rate
Agentic business operations
Quantitative analysis on spreadsheets & documents
Instruction following
Long-horizon agentic tasks
Kubernetes incident root-cause analysis
Visual reasoning
Intelligence Index vs. Price
Context Window
Context Window
Pricing
Intelligence Index vs. Price
Performance Summary
Output Speed vs. Price
Speed
Measured by Output Speed (tokens per second)
Output Speed
Latency
Measured by Time (seconds) to First Token
Latency: Time To First Answer Token
End-to-End Response Time
Seconds to output 500 tokens, calculated based on time to first token, 'thinking' time for reasoning models, and output speed
End-to-End Response Time vs. Price
Further Analysis | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
GLM-5.2 (max) | 262k | Open | 53 | $0.38 | 140 | 1.43 | 19.33 | 14.32 | |||
DeepSeek V4 Flash 0731 (max) | 262k | Open | 52 | $0.12 | 158 | 1.66 | 17.46 | 12.64 | |||
MiniMax-M3 | 262k | Open | 45 | $0.11 | 121 | 1.37 | 22.04 | 16.53 | |||
Kimi K2.6 | 262k | Open | 45 | -- | 202 | 1.14 | 25.64 | 22.03 | |||
Kimi K2.7 Code | 262k | Open | 43 | $0.28 | 293 | 1.20 | 10.50 | 7.60 | |||
GLM-5.1 | 203k | Open | 41 | $0.58 | 206 | 1.25 | 22.10 | 18.42 | |||
GLM-5 (FP8) | 200k | Open | 41* | -- | -- | -- | -- | -- | |||
DeepSeek V4 Flash (high) | 1M | Open | 39 | $0.07 | 64 | 1.57 | 28.71 | 19.34 | |||
Nemotron 3 Ultra | 262k | Open | 38 | $0.16 | 226 | 1.02 | 13.32 | 10.09 | |||
Kimi K2.6 | 262k | Open | 35* | -- | 192 | 1.11 | 3.71 | -- | |||
MiniMax-M2.5 | 197k | Open | 34* | -- | 84 | 1.09 | 30.90 | 23.85 | |||
 Gemma 4 31B | 262k | Open | 30 | -- | 36 | 1.91 | 63.97 | 48.19 | |||
DeepSeek V4 Flash | 1M | Open | 29* | -- | 64 | 1.59 | 9.44 | -- | |||
Nemotron 3 Super | 262k | Open | 26 | $0.23 | 143 | 0.95 | 18.38 | 13.95 | |||
gpt-oss-120b (high) | 131k | Open | 24 | $0.02 | 34 | 1.51 | 74.35 | 58.27 | |||
Nemotron 3.5 Lightning (BF16) | 262k | Open | 24 | $0.08 | 343 | 0.86 | 8.15 | 5.83 | |||
DeepSeek V3.1 | 128k | Open | 21* | -- | 62 | 1.39 | 9.49 | -- | |||
 Qwen3 Coder 480B | 262k | Open | 18* | -- | 67 | 1.20 | 8.63 | -- | |||
gpt-oss-20b (high) | 131k | Open | 15 | $0.01 | 90 | 0.90 | 28.60 | 22.17 | |||
gpt-oss-120b (low) | 131k | Open | 15 | $0.0042 | 33 | 1.51 | 78.11 | 61.28 | |||
gpt-oss-20b (low) | 131k | Open | 14* | -- | 94 | 0.95 | 27.50 | 21.24 | |||
 Llama 3.3 70B | 128k | Open | 9* | -- | 79 | 0.89 | 7.24 | -- | |||
Qwen3 30B A3B 2507 | 262k | Open | 9* | -- | 143 | 0.95 | 4.45 | -- | |||
Llama 3.1 8B | 128k | Open | 7* | -- | 138 | 0.99 | 4.62 | -- | |||
Granite 4.1 8B | 131k | Open | 6* | -- | 113 | 0.77 | 5.18 | -- | |||
Key definitions
Frequently Asked Questions
Common questions about CoreWeave
CoreWeave offers 24 models that we track: GLM-5.2 (max), DeepSeek V4 Flash 0731 (max), MiniMax-M3, Kimi K2.6, Kimi K2.7 Code, GLM-5.1, DeepSeek V4 Flash (high), Nemotron 3 Ultra, Kimi K2.6, MiniMax-M2.5, Gemma 4 31B, DeepSeek V4 Flash, Nemotron 3 Super, gpt-oss-120b (high), Nemotron 3.5 Lightning (BF16), DeepSeek V3.1, Qwen3 Coder 480B, gpt-oss-20b (high), gpt-oss-120b (low), gpt-oss-20b (low), Llama 3.3 70B, Qwen3 30B A3B 2507, Llama 3.1 8B, and Granite 4.1 8B.
The most intelligent model available on CoreWeave is GLM-5.2 (max) with an Intelligence Index score of 53.
The fastest model on CoreWeave by output speed is Nemotron 3.5 Lightning (BF16) at 342.8 tokens per second.
The model with the lowest time to first answer token on CoreWeave is Granite 4.1 8B at 0.77s. Lower latency means faster initial response time.
The most affordable model on CoreWeave by blended price is gpt-oss-20b (high) at $0.04 per 1M tokens (7:2:1 cache hit/input/output ratio).
Prices on CoreWeave vary up to 43x across models, from $0.04 per 1M tokens for gpt-oss-20b (high) to $1.70 per 1M tokens for GLM-5.1.
Yes, CoreWeave offers an OpenAI-compatible API, making it easy to switch from OpenAI or use existing OpenAI SDK integrations.
Yes, all 24 models on CoreWeave support JSON mode for structured output.
Yes, all 24 models on CoreWeave support function calling (tool use).
Yes, CoreWeave offers 16 reasoning models: GLM-5.2 (max), DeepSeek V4 Flash 0731 (max), MiniMax-M3, Kimi K2.6, Kimi K2.7 Code, GLM-5.1, DeepSeek V4 Flash (high), Nemotron 3 Ultra, MiniMax-M2.5, Gemma 4 31B, Nemotron 3 Super, gpt-oss-120b (high), Nemotron 3.5 Lightning (BF16), gpt-oss-20b (high), gpt-oss-120b (low), and gpt-oss-20b (low). Reasoning models use extended thinking to work through complex problems before providing an answer.
Yes, all 24 models on CoreWeave are open weight models.
Yes, provider performance can vary over time due to infrastructure changes, load balancing, and updates. We continuously benchmark all providers and display historical performance trends in the "Over Time" charts.
When choosing a model on CoreWeave, consider: intelligence (for quality-sensitive tasks), output speed (for throughput-intensive tasks), latency (for interactive applications requiring quick first responses), pricing (for cost-sensitive workloads), and features like context window size, JSON mode, or function calling support.