Fireworks AI · none caching
Devstral-Small-2505
Devstral-Small-2505 on Zumik: live pricing, context, and caching, routable by id or alias through one OpenAI-compatible endpoint.
Specifications
At a glance.
| Provider | Fireworks AI |
| Family | mistral |
| Released | 2025-05 |
| License | Open weights |
| Context window | 128K tokens |
| Max output | 128K tokens |
| Parameters | 24B |
| Modalities | text |
| Tool calling | Yes |
| Reasoning mode | No |
| Caching | none |
| Batch discount | No batch tier |
Measured by Zumik
What reuse looks like here.
Pricing, context, and capabilities for Devstral-Small-2505 are live, but it is outside the flagship set Zumik benchmarks in depth, so measured reuse, capture, and warm TTFT are not shown yet. Run a workload estimate or route it by id to start collecting traces.
What you actually pay once caching works.
At a typical 55% prefix reuse, a million input tokens on Devstral-Small-2505 effectively costs $0.10 instead of $0.10 - blending to roughly $0.15 with a 25% output share. There is no batch tier, so cost control here leans on caching and routing.
Estimate it for your workloadRoute it directly by id, or let an alias pick it when it wins under policy.
Call it
Same OpenAI client, this model.
from openai import OpenAI
client = OpenAI(base_url="https://api.zumik.ai/v1", api_key="zk_live_...")
r = client.responses.create(
model="devstral-small-2505",
input="Draft a fix for the failing test.",
)
print(r.usage.input_tokens_cached) # confirm reuseFrequently asked
Devstral-Small-2505, answered.
How much does Devstral-Small-2505 cost?
Devstral-Small-2505 is $0.10 per million input tokens and $0.30 per million output tokens through Zumik. Cache reads are $0.10 per million, a 0% discount on input.
What is Devstral-Small-2505's context window?
Devstral-Small-2505 supports a 128K-token context window with up to 128K output tokens.
Does Devstral-Small-2505 support prompt caching?
Yes. Fireworks AI uses Automatic prompt caching (serverless and dedicated) caching. In the Zumik corpus, Devstral-Small-2505 shows a median cache capture of 76% on agent workloads.
Run Devstral-Small-2505 with reuse measured.
Point an OpenAI client at Zumik and see exactly how much of this model's input you are reusing.