Particle-1
Particle AI post-trained DeepSeek V4.1 Flash for coding tasks with less generated text. Its evaluation pairs task pass rates with output-token use so you can assess both quality and efficiency.
Particle-1 is available on MiniRouter as . The model page has current pricing, provider routes and API setup.
Compare useful work and generated tokens together.
Particle’s DeepSWE 1.1 evaluation pairs a similar task pass rate with lower output volume than its DeepSeek V4.1 Flash base. The result is a reason to test your own workload, rather than a guarantee of lower cost.
DeepSWE 1.1 task pass rate
- Particle-172.6%
- DeepSeek V4.1 Flash71.2%
0–100% · higher is better
- Particle-1
- 59,580
- mean output tokens per task
- DeepSeek V4.1 Flash
- 83,246
- mean output tokens per task
View chart data
| Model | Task pass rate | Mean output tokens |
|---|---|---|
| Particle-1 | 72.57% | 59,580 |
| DeepSeek V4.1 Flash | 71.15% | 83,246 |
Try a coding task with an acceptance check.
- Keep the prompt, tools and tests fixed when comparing models.
- Measure completed tasks, retries and total spend, including reasoning tokens.
- Inspect whether shorter runs retain the tests and edge cases you need.
Try Particle-1
Create a MiniRouter key and supply the relevant code with an explicit output cap. This route accepts text input.
curl https://api.minirouter.sh/v1/chat/completions \
-H "Authorization: Bearer $MINIROUTER_API_KEY" \
-H 'Content-Type: application/json' \
--data '{
"model": "particle/particle-1",
"max_tokens": 2048,
"messages": [{ "role": "user", "content": "Review the supplied patch, identify missing tests and suggest a minimal fix." }]
}'Keep exploring
Compare published results in the model rankings and coding benchmarks, or read more model releases.
For MiniRouter product updates, visit the changelog.
Sources
- Read at Particle AI ↗
Particle-1 announcement
Published Oct 28, 2026
- Read at Particle AI ↗
Particle model documentation
Published Oct 28, 2026
Editorial reference revised 2026-10-29. Catalog status comes from the generated routing snapshot.