Fireworks AI API pricing
Updated
Fireworks AI lists standard per-token prices for 15 models. The cheapest is NVIDIA Nemotron 3.5 Lightning 30B A3B at USD 0.05 input and USD 0.20 output per 1M tokens; the dearest is Kimi K3 at USD 3 / USD 15. It publishes cached-input prices for all of them, plus flex / spot, priority and fast tiers (no batch). Checked 2026-11-14.
- Models with a standard price
- 15
- Median input / output
- USD 0.30 / USD 1.20
- Cached-input prices
- 15 of 15
- Cheapest host for
- 0 of 13 shared models
Which models is Fireworks AI the cheapest provider for?
None outright. Of the 13 models that Fireworks AI shares with other providers, 11 are cheaper elsewhere at the standard tier, and Fireworks AI ties the lowest price for the other 2, where another provider charges the same. It is the dearest provider for 4.
How does Fireworks AI price the most widely hosted models?
The 6 models Fireworks AI sells that the most other providers also sell, with Fireworks AI's rank among them on the 3:1 blend.
| Model | Fireworks AI | Rank | Cheapest |
|---|---|---|---|
| GLM 5.3 | USD 1.40 / USD 4.40 | 6 of 24 | DeepInfra USD 0.5625 / USD 2.50 |
| OpenAI GPT OSS 120B | USD 0.15 / USD 0.60 | 11 of 21 | CoreWeave Forge USD 0.03 / USD 0.17 |
| GLM 5.3 Flash | USD 0.15 / USD 0.50 | 3 of 20 | GMI Cloud USD 0.075 / USD 0.25 |
| Kimi K3 | USD 3 / USD 15 | 5 of 18 | Chutes USD 1.85 / USD 9.50 |
| DeepSeek V4.1 Flash | USD 0.30 / USD 1.20 | 5 of 16 | DeepInfra USD 0.14 / USD 0.42 |
| DeepSeek V4.1 Flash 1017 | USD 0.30 / USD 1.20 | 6 of 16 | DeepInfra USD 0.098 / USD 0.294 |
Which model makers does Fireworks AI sell?
Fireworks AI prices models from 9 makers. DeepSeek leads with 2 models. Standard prices, USD per 1M input / output tokens.
| Maker | Models | Cheapest model | Median input / output |
|---|---|---|---|
| DeepSeek | 2 | DeepSeek V4.1 Flash USD 0.30 / USD 1.20 | USD 0.30 / USD 1.20 |
| Fireworks AI | 2 | Ultramarine 1 Flash USD 0.15 / USD 0.50 | USD 1.575 / USD 7.75 |
| MiniMax | 2 | MiniMax M3.1 Flash USD 0.15 / USD 0.60 | USD 0.225 / USD 0.90 |
| Moonshot AI | 2 | Kimi K3.1 Pro USD 2 / USD 10 | USD 2.50 / USD 12.50 |
| NVIDIA | 2 | NVIDIA Nemotron 3.5 Lightning 30B A3B USD 0.05 / USD 0.20 | USD 0.325 / USD 1.30 |
| Z.ai | 2 | GLM 5.3 Flash USD 0.15 / USD 0.50 | USD 0.775 / USD 2.45 |
| Meta | 1 | Muse Spark Open USD 1.25 / USD 4.25 | — |
| OpenAI | 1 | OpenAI GPT OSS 120B USD 0.15 / USD 0.60 | — |
| Qwen | 1 | Qwen3.8 Max USD 2 / USD 6 | — |
Fireworks AI price list
USD per 1M tokens: 36 prices for 15 models, one row per model with any other service tiers stacked under the standard price. Sources checked 2026-11-14 unless a row says otherwise. "vs maker" compares a standard price with the model maker's own API on the 3:1 blend; a minus means Fireworks AI is cheaper. Of 9 hosted model prices with a maker price to compare, 0 are below the maker's and 0 above.
| Model | Tier | Input | Cached | Output | Context | vs maker | Source |
|---|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash | standard flex / spot priority | USD 0.30 USD 0.21 USD 0.375 | USD 0.006 USD 0.0042 USD 0.0075 | USD 1.20 USD 0.84 USD 1.50 | 1,048,576 | same | source source 2 |
| DeepSeek V4.1 Flash 1017 | standard flex / spot priority fast | USD 0.30 USD 0.21 USD 0.375 USD 2.40 | USD 0.006 USD 0.0042 USD 0.0075 USD 0.048 | USD 1.20 USD 0.84 USD 1.50 USD 9.60 | 1,048,576 | same | source source 2 |
| Ember-1 | standard priority | USD 3 USD 3.75 | USD 0.30 USD 0.375 | USD 15 USD 18.75 | 1,048,576 | — | source |
| GLM 5.3 | standard flex / spot priority fast | USD 1.40 USD 0.98 USD 1.75 USD 2.10 | USD 0.26 USD 0.182 USD 0.325 USD 0.39 | USD 4.40 USD 3.08 USD 5.50 USD 6.60 | 1,048,576 | same | source source 2 |
| GLM 5.3 Flash | standard priority | USD 0.15 USD 0.1875 | USD 0.03 USD 0.0375 | USD 0.50 USD 0.625 | 1,048,576 | same | source |
| Kimi K3 | standard flex / spot priority fast | USD 3 USD 2.10 USD 3.75 USD 4.50 | USD 0.30 USD 0.21 USD 0.375 USD 0.45 | USD 15 USD 10.50 USD 18.75 USD 22.50 | 1,048,576 | same | source source 2 |
| Kimi K3.1 Pro | standard priority | USD 2 USD 3 | USD 0.20 USD 0.30 | USD 10 USD 15 | 1,048,576 | same | source |
| MiniMax M3 | standard priority | USD 0.30 USD 0.45 | USD 0.06 USD 0.09 | USD 1.20 USD 1.80 | 512,000 | same | source |
| MiniMax M3.1 Flash | standard | USD 0.15 | USD 0.015 | USD 0.60 | 1,048,576 | same | source |
| Muse Spark Open | standard priority | USD 1.25 USD 1.875 | USD 0.15 USD 0.225 | USD 4.25 USD 6.375 | 1,048,576 | — | source |
| NVIDIA Nemotron 3 Ultra | standard priority | USD 0.60 USD 0.75 | USD 0.12 USD 0.15 | USD 2.40 USD 3 | not stated | — | source |
| NVIDIA Nemotron 3.5 Lightning 30B A3B | standard priority | USD 0.05 USD 0.0625 | USD 0.01 USD 0.0125 | USD 0.20 USD 0.25 | 262,144 | — | source |
| OpenAI GPT OSS 120B | standard priority | USD 0.15 USD 0.18 | USD 0.015 USD 0.018 | USD 0.60 USD 0.72 | not stated | — | source |
| Qwen3.8 Max | standard priority | USD 2 USD 3 | USD 0.25 USD 0.375 | USD 6 USD 9 | not stated | same | source |
| Ultramarine 1 Flash | standard priority | USD 0.15 USD 0.1875 | USD 0.03 USD 0.0375 | USD 0.50 USD 0.625 | 1,048,576 | — | source |
- Cached "—" means Fireworks AI's page shows no cache price for that row; budget cached tokens at the input price.
- Context "not stated": the source does not give it (30 of 36 rows have one).
Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
What service tiers does Fireworks AI publish?
- flex / spot (4 rows). Across 4 models with both prices, the median flex / spot price is 30% below standard. Fireworks AI row note: “Fireworks "Spot" mode: 30% off Standard on best-effort capacity; throughput can drop to zero.”
- priority (14 rows). Across 14 models with both prices, the median priority price is 1.25× standard (range 1.2× standard to 1.5× standard). Fireworks AI row note: “Priority mode.”
- fast (3 rows). Across 3 models with both prices, the median fast price is 1.5× standard (range 1.5× standard to 8× standard). Fireworks AI row note: “Fast mode.”
Cached input: 15 of 15 standard prices include a cache-read price, a median 12% of the input price. What cached input means.
What we recorded about Fireworks AI's pricing page: Per-model rates from the serverless pricing docs (linked from fireworks.ai/pricing as "See all models"). US-only (+50%) rows skipped. Batch is published only as a blanket "50% off" rule, so no batch rows. Spot rows use tier "flex". Models not listed individually are priced by size (from $0.10 to $1.20 per 1M tokens) and are not included. Rune 26B-A4B (decision model, no output tokens), embeddings and rerankers excluded.
Is Fireworks AI cheaper than the model makers' own APIs?
Mostly it charges the same as the maker. Of 9 standard prices for models whose maker sells its own API, 0 are below the maker's price, 0 above and 9 the same, on the 3:1 blend. The median difference is zero.
BroadNet Inference and Fireworks AI
BroadRouter is BroadNet's own site, so BroadNet is never ranked on this page. BroadNet Inference also serves 2 models that Fireworks AI prices: DeepSeek V4.1 Flash at USD 0.175 / USD 0.52 against Fireworks AI's USD 0.30 / USD 1.20 (BroadNet is cheaper on the 3:1 blend); GLM 5.3 at USD 0.78 / USD 3.46 against Fireworks AI's USD 1.40 / USD 4.40 (BroadNet is cheaper on the 3:1 blend). BroadNet prices are from its rate card, synced 2026-11-14. BroadNet Inference pricing.
More on Fireworks AI
- Fireworks AI's own pricing page (the source; checked 2026-11-14)
- Fireworks AI alternative: when to switch and what to use
- LLM API free tiers and rate limits: Fireworks AI's free tier and default limits next to other providers'
- All 41 providers compared · Cheapest LLM APIs right now · Download every price (CSV, JSON)
Quick answers
Is Fireworks AI OpenAI-compatible?
Yes, for chat completions. Model IDs take the form accounts/fireworks/models/..., and Fireworks lowers max_tokens to fit the context instead of returning an error, unless told otherwise (Fireworks AI, OpenAI compatibility, checked 2026-11-15).
What is Fireworks AI's API base URL?
https://api.fireworks.ai/inference/v1, as Fireworks AI documents it (Fireworks AI, OpenAI compatibility, checked 2026-11-15).
Does Fireworks AI have a free tier?
No free credit mentioned on the rate-limits page. Default limits (No payment method or no credits, Account-wide): 10 requests a minute. 10 RPM without a payment method or credits; up to 6,000 RPM with a payment method and active credits (Fireworks AI rate limits docs, checked 2026-11-14).
Does Fireworks AI have a batch API?
Yes. Fireworks lists a Batch mode at 50% off as a blanket rule, with no per-model batch prices (Fireworks AI pricing, checked 2026-11-14).
Questions
How much does the Fireworks AI API cost per token?
Fireworks AI's standard prices run from USD 0.05 input and USD 0.20 output per 1M tokens (NVIDIA Nemotron 3.5 Lightning 30B A3B) to USD 3 / USD 15 (Kimi K3). The median across 15 standard prices is USD 0.30 input and USD 1.20 output per 1M, which is USD 0.0000003 per input token. Input, cached input and output are billed separately.
What is the cheapest model on Fireworks AI?
NVIDIA Nemotron 3.5 Lightning 30B A3B, at USD 0.05 input and USD 0.20 output per 1M tokens (USD 0.0875 on a 3:1 input-to-output blend). The cheapest from other makers: DeepSeek V4.1 Flash (DeepSeek) at USD 0.30 / USD 1.20; Ultramarine 1 Flash (Fireworks AI) at USD 0.15 / USD 0.50; MiniMax M3.1 Flash (MiniMax) at USD 0.15 / USD 0.60.
Is Fireworks AI cheaper than other providers?
It depends on the model. Of 13 models that other providers also price, Fireworks AI is the cheapest standard price for none; it ties the lowest price, without beating it, for 2; another provider is cheaper for 11; and it is the dearest for 4. Its median rank is 4 of a median 13 providers, on a 3:1 input-to-output blend.
Does Fireworks AI have batch, cached or priority pricing?
Cached input: yes, for 15 of 15 standard prices, at a median 12% of the input price. Other tiers: flex / spot on 4 rows (median input 30% below standard); priority on 14 rows (median input 1.25× standard); fast on 3 rows (median input 1.5× standard).
How much does a chatbot cost per month on Fireworks AI?
For 2,000 requests a day, 1,500 input tokens (60% cached), 300 output, over 30 days: USD 5.94 on NVIDIA Nemotron 3.5 Lightning 30B A3B, USD 35.64 on MiniMax M3 (the middle of Fireworks AI's list) and USD 394.20 on Kimi K3. Try your own numbers.
Read as MarkdownAsk ChatGPTAsk ClaudeAsk Perplexity
Sources