Fireworks: Models Intelligence, Performance & Price
This analysis is intended to support you in choosing the best model provided by Fireworks for your use-case.
Most Intelligent
Intelligence Index
Total 15 models
Lowest Time per Task
Time per Intelligence Index task (minutes)
Total 15 models
Lowest Price
Blended price (per 1M tokens)
Total 15 models
Fireworks offers 15 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 Fireworks are Kimi K3.1 Pro (46), DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST) (33), and DeepSeek V4.1 Flash 1017 (max) (33).
- For time per task, DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST) (3.2 min), Nemotron 3 Ultra (4.3 min), and MiniMax-M3 (5.8 min) finish an Intelligence Index task fastest on Fireworks. Time per task varies significantly across models, with the slowest taking 2.3x longer than the quickest.
- For latency, DeepSeek V4.1 Flash (non-reasoning) (0.94s), Nemotron 3 Ultra (3.46s), and MiniMax-M3 (4.10s) offer the lowest time to first answer token.
- For pricing, MiniMax-M3.1-Flash ($0.10), GLM-5.3-Flash ($0.10), and DeepSeek V4.1 Flash 1017 (max) ($0.18) offer the lowest blended prices per 1M tokens.
- For context window size, Kimi K3.1 Pro (1M), DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST) (1M), and GLM-5.3 (max) (1M) support the largest context windows on Fireworks.
Intelligence
Artificial Analysis Intelligence Index
Evaluations
Agentic knowledge work, Elo
Knowledge
1 - hallucination rate
Agentic SaaS workflows
Agentic scientific research workflows in a terminal
Agentic software engineering in open-source repositories
Reasoning & knowledge
Professional document reasoning, All-pass
Agentic coding & terminal use
Physics reasoning
Quantitative analysis on spreadsheets & documents
Legal agentic work, Hallucination-Gated All-Pass Rate
Agentic business operations
Agentic real-world work tasks, Elo
Agentic interactive 3D experiences, Elo
Product and brand videos designed in code, Elo
Kubernetes incident root-cause analysis
Medical long context reasoning
Visual reasoning
Rich, varied, well-crafted language
Well-structured, scannable, answer-first
Truthful, acknowledges uncertainty, non-sycophantic
Use of filler and generic language
Intelligence Index vs. Price
Pricing
Cost per Intelligence Index Task
Intelligence Index vs. Price
Speed
Average of the single-request and multi-turn agentic workloads measured
Time per Intelligence Index Task
Quality
Endpoint Accuracy Index
Latency
Average of the single-request and multi-turn agentic workloads measured
Time to First Answer Token
Reliability
Share of requests the endpoint failed, across all the single-request and multi-turn agentic workloads measured
Endpoint Error Rate
Caching
Cache Hit Rate
Cost per Task vs. Cache Hit Rate
Further Analysis | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Kimi K3.1 Pro | 46 | $5.24 | 19.1 | 82.12 | 0.7% | 1M | Open | |||
DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST) | 33 | $21.80 | 3.8 | 13.89 | 0.0% | 1M | Open | |||
DeepSeek V4.1 Flash 1017 (max) | 33 | $1.22 | 10.4 | 31.90 | 0.8% | 1M | Open | |||
GLM-5.3 (max) | 32 | $5.41 | 17.3 | 50.50 | 0.0% | 1M | Open | |||
GLM-5.3 (max) (FAST) | 32 | $10.32 | 9.0 | 20.96 | 4.7% | 1M | Open | |||
MiniMax-M3.1-Flash | 29 | $0.50 | 16.6 | 33.21 | 2.2% | 1M | Open | |||
GLM-5.3-Flash | 28 | $0.78 | 14.5 | 20.93 | 0.0% | 1M | Open | |||
Muse Spark Open (xhigh) | 25 | $2.02 | 7.6 | 11.85 | 0.0% | 1M | Open | |||
MiniMax-M3 | 19 | $1.26 | 6.5 | 5.72 | 0.0% | 524k | Open | |||
Kimi K3 (low) | 17 | $1.94 | 7.8 | 4.71 | 0.0% | 1M | Open | |||
Nemotron 3 Ultra | 12 | $2.21 | 4.3 | 3.03 | 0.8% | 262k | Open | |||
Kimi K2.6 | -- | -- | -- | -- | -- | 262k | Open | |||
Kimi K2.6 (non-reasoning) | -- | -- | -- | -- | -- | 262k | Open | |||
Qwen3 30B | -- | -- | -- | -- | -- | 262k | Open | |||
GLM-5 | -- | -- | -- | -- | -- | 203k | Open | |||
Qwen3 VL 30B A3B | -- | -- | -- | -- | -- | 262k | Open | |||
DeepSeek V3.2 (non-reasoning) | -- | -- | -- | -- | -- | 164k | Open | |||
DeepSeek V4.1 Flash (non-reasoning) | -- | -- | -- | 1.35 | 0.0% | 1M | Open | |||
Qwen3.8 2.4T A95B | -- | -- | -- | 27.87 | 1.5% | 262k | Open | |||
Qwen3 8B (non-reasoning) | -- | -- | -- | -- | -- | 41k | Open | |||
Qwen3 VL 30B A3B | -- | -- | -- | -- | -- | 262k | Open | |||
Qwen3 VL 235B A22B | -- | -- | -- | -- | -- | 262k | Open | |||
DeepSeek V3.2 | -- | -- | -- | -- | -- | 164k | Open | |||
DeepSeek V4.1 Flash (max) | -- | -- | -- | 81.62 | 0.0% | 1M | Open | |||
GLM-5 (non-reasoning) | -- | -- | -- | -- | -- | 203k | Open | |||
GLM-5.2 (max) | -- | -- | -- | -- | -- | 1M | Open | |||
Kimi K3 (max) | -- | -- | -- | 23.57 | 0.0% | 1M | Open | |||
DeepSeek V3.1 (non-reasoning) | -- | -- | -- | -- | -- | 164k | Open | |||
DeepSeek V4 Flash 0731 (max) | -- | -- | -- | -- | -- | 1M | Open | |||
Key definitions
Frequently Asked Questions
Common questions about Fireworks
Fireworks offers 15 models that we track: Kimi K3.1 Pro, DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST), DeepSeek V4.1 Flash 1017 (max), GLM-5.3 (max), GLM-5.3 (max) (FAST), MiniMax-M3.1-Flash, GLM-5.3-Flash, Muse Spark Open (xhigh), MiniMax-M3, Kimi K3 (low), Nemotron 3 Ultra, DeepSeek V4.1 Flash (non-reasoning), Qwen3.8 2.4T A95B, DeepSeek V4.1 Flash (max), and Kimi K3 (max).
The most intelligent model available on Fireworks is Kimi K3.1 Pro with an Intelligence Index score of 46.
The model with the lowest time per task on Fireworks is DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST) at 3.2 minutes per Intelligence Index task. Time per task is the output tokens per task divided by end-to-end output speed, so lower means a task finishes sooner.
The model with the lowest time to first answer token on Fireworks is DeepSeek V4.1 Flash (non-reasoning) at 0.94s. Lower latency means faster initial response time.
The most affordable model on Fireworks by blended price is MiniMax-M3.1-Flash at $0.10 per 1M tokens (7:2:1 cache hit/input/output ratio).
Prices on Fireworks vary up to 23.1x across models, from $0.10 per 1M tokens for MiniMax-M3.1-Flash to $2.31 per 1M tokens for Kimi K3 (max).
Yes, Fireworks offers an OpenAI-compatible API, making it easy to switch from OpenAI or use existing OpenAI SDK integrations.
Yes, all 15 models on Fireworks support JSON mode for structured output.
Yes, all 15 models on Fireworks support function calling (tool use).
Yes, Fireworks offers 14 reasoning models: Kimi K3.1 Pro, DeepSeek V4.1 Flash 1017 (max) (ULTRA FAST), DeepSeek V4.1 Flash 1017 (max), GLM-5.3 (max), GLM-5.3 (max) (FAST), MiniMax-M3.1-Flash, GLM-5.3-Flash, Muse Spark Open (xhigh), MiniMax-M3, Kimi K3 (low), Nemotron 3 Ultra, Qwen3.8 2.4T A95B, DeepSeek V4.1 Flash (max), and Kimi K3 (max). Reasoning models use extended thinking to work through complex problems before providing an answer.
Yes, all 15 models on Fireworks 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 Fireworks, consider: intelligence (for quality-sensitive tasks), time per task (how quickly the model completes an Intelligence Index task), 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.