Parasail: Models Intelligence, Performance & Price
Analysis of Parasail'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 Parasail for your use-case.
Most Intelligent
Intelligence index
Total 29 models
Fastest
Output speed
Total 29 models
Lowest Price
Blended price (per 1M tokens)
Total 29 models
Parasail offers 29 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 Parasail are Kimi K3 (max) (60), GLM-5.2 (max) (NVFP4) (53), and DeepSeek V4 Flash 0731 (max) (52).
- For output speed, the fastest models are Trinity Large Thinking (FP8) (193 t/s), Kimi K3 (max) (186 t/s), and Kimi K2.6 (154 t/s).
- For latency, Qwen3.6 35B A3B (FP8) (0.95s), Qwen3 Next 80B A3B (0.97s), and Llama 4 Maverick (FP8) (1.11s) offer the lowest time to first answer token.
- For pricing, Gemma 3 27B ($0.09), MiMo-V2.5 ($0.09), and Gemma 4 26B A4B ($0.10) offer the lowest blended prices per 1M tokens.
- For context window size, Kimi K3 (max) (1M), DeepSeek V4 Flash 0731 (max) (1M), and DeepSeek V4 Flash (max) (FP8) (1M) support the largest context windows on Parasail.
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
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 | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
Kimi K3 (max) | 1.05M | Open | 60 | $0.78 | 186 | 2.01 | 15.45 | 10.76 | |||
GLM-5.2 (max) (NVFP4) | 1M | Open | 53 | -- | 145 | 1.23 | 18.42 | 13.76 | |||
DeepSeek V4 Flash 0731 (max) | 1.05M | Open | 52 | $0.13 | 107 | 1.27 | 24.66 | 18.71 | |||
MiniMax-M3 (MXFP8) | 1M | Open | 45 | $0.18 | 96 | 1.32 | 27.45 | 20.91 | |||
Kimi K2.6 | 262k | Open | 45 | $0.27 | 154 | 1.50 | 33.71 | 28.96 | |||
Kimi K2.7 Code | 262k | Open | 43 | $0.18 | 88 | 1.22 | 32.29 | 25.38 | |||
DeepSeek V4 Flash (max) (FP8) | 1.05M | Open | 42 | -- | 85 | 1.58 | 73.76 | 66.28 | |||
GLM-5.1 (FP8) | 203k | Open | 41 | $0.24 | 83 | 1.77 | 53.31 | 45.53 | |||
DeepSeek V4 Flash (high) (FP8) | 1.05M | Open | 39 | -- | 88 | 1.46 | 21.27 | 14.12 | |||
MiMo-V2.5 | 1M | Open | 38 | $0.04 | 76 | 1.01 | 33.75 | 26.19 | |||
GLM-5.1 | 203k | Open | 36* | -- | 89 | 1.65 | 7.28 | -- | |||
Kimi K2.6 (INT4) | 262k | Open | 35* | -- | 123 | 1.55 | 5.62 | -- | |||
 Qwen3.5 397B A17B | 262k | Open | 34 | $0.23 | 89 | 1.02 | 42.36 | 35.74 | |||
Qwen3.6 35B A3B | 262k | Open | 32 | $0.08 | 73 | 1.32 | 82.37 | 74.18 | |||
Gemma 4 31B | 262k | Open | 30 | $0.02 | 13 | 3.96 | 181.27 | 137.66 | |||
Gemma 4 26B A4B | 256k | Open | 26 | $0.03 | 30 | 2.06 | 84.71 | 66.12 | |||
Qwen3.6 35B A3B (FP8) | 262k | Open | 25 | $0.15 | 90 | 0.95 | 6.48 | -- | |||
 gpt-oss-120b (high) | 131k | Open | 24 | -- | 101 | 1.02 | 25.88 | 19.89 | |||
Gemma 4 31B | 262k | Open | 22 | $0.03 | 12 | 4.01 | 46.58 | -- | |||
Qwen3 Coder Next (FP8) | 262k | Open | 21 | $0.08 | 59 | 1.13 | 9.63 | -- | |||
Gemma 4 26B A4B | 262k | Open | 20* | -- | 35 | 1.89 | 16.06 | -- | |||
Trinity Large Thinking (FP8) | 262k | Open | 19 | $0.08 | 193 | 0.89 | 13.86 | 10.38 | |||
Qwen3 235B 2507 | 262k | Open | 18* | -- | 34 | 1.22 | 16.03 | -- | |||
gpt-oss-120b (low) | 131k | Open | 15 | -- | 93 | 0.98 | 27.93 | 21.57 | |||
Llama 4 Maverick (FP8) | 1.05M | Open | 14 | $0.03 | 145 | 1.11 | 4.55 | -- | |||
Qwen3 VL 235B A22B (FP8) | 131k | Open | 14* | -- | 42 | 1.12 | 13.14 | -- | |||
Qwen3 Next 80B A3B | 262k | Open | 14* | -- | 137 | 0.97 | 4.62 | -- | |||
Llama 3.3 70B (FP8) | 131k | Open | 9 | $0.02 | 49 | 2.58 | 12.71 | -- | |||
 Olmo 3.1 32B Think | 65.5k | Open | 8* | -- | -- | -- | -- | -- | |||
Gemma 3 27B | 131k | Open | 7 | $0.05 | 76 | 1.19 | 7.80 | -- | |||
Olmo 3 7B | 65.5k | Open | 2* | -- | -- | -- | -- | -- | |||
Key definitions
Frequently Asked Questions
Common questions about Parasail
Parasail offers 29 models that we track: Kimi K3 (max), GLM-5.2 (max) (NVFP4), DeepSeek V4 Flash 0731 (max), MiniMax-M3 (MXFP8), Kimi K2.6, Kimi K2.7 Code, DeepSeek V4 Flash (max) (FP8), GLM-5.1 (FP8), DeepSeek V4 Flash (high) (FP8), MiMo-V2.5, GLM-5.1, Kimi K2.6 (INT4), Qwen3.5 397B A17B, Qwen3.6 35B A3B, Gemma 4 31B, Gemma 4 26B A4B, Qwen3.6 35B A3B (FP8), gpt-oss-120b (high), Gemma 4 31B, Qwen3 Coder Next (FP8), Gemma 4 26B A4B, Trinity Large Thinking (FP8), Qwen3 235B 2507, gpt-oss-120b (low), Llama 4 Maverick (FP8), Qwen3 VL 235B A22B (FP8), Qwen3 Next 80B A3B, Llama 3.3 70B (FP8), and Gemma 3 27B.
The most intelligent model available on Parasail is Kimi K3 (max) with an Intelligence Index score of 60.
The fastest model on Parasail by output speed is Trinity Large Thinking (FP8) at 192.7 tokens per second.
The model with the lowest time to first answer token on Parasail is Qwen3.6 35B A3B (FP8) at 0.95s. Lower latency means faster initial response time.
The most affordable model on Parasail by blended price is Gemma 3 27B at $0.09 per 1M tokens (7:2:1 cache hit/input/output ratio).
Prices on Parasail vary up to 26x across models, from $0.09 per 1M tokens for Gemma 3 27B to $2.31 per 1M tokens for Kimi K3 (max).
Yes, Parasail offers an OpenAI-compatible API, making it easy to switch from OpenAI or use existing OpenAI SDK integrations.
Yes, all 29 models on Parasail support JSON mode for structured output.
Yes, all 29 models on Parasail support function calling (tool use).
Yes, Parasail offers 17 reasoning models: Kimi K3 (max), GLM-5.2 (max) (NVFP4), DeepSeek V4 Flash 0731 (max), MiniMax-M3 (MXFP8), Kimi K2.6, Kimi K2.7 Code, DeepSeek V4 Flash (max) (FP8), GLM-5.1 (FP8), DeepSeek V4 Flash (high) (FP8), MiMo-V2.5, Qwen3.5 397B A17B, Qwen3.6 35B A3B, Gemma 4 31B, Gemma 4 26B A4B, gpt-oss-120b (high), Trinity Large Thinking (FP8), and gpt-oss-120b (low). Reasoning models use extended thinking to work through complex problems before providing an answer.
Yes, all 29 models on Parasail 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 Parasail, 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.