Primary use stated or inferred cautiously from official documentation.
I
INCLUSIONAI / inclusionAI/Ling-3.0-flash-Sante-fp8
Ling 3.0 Flash Sante FP8
An FP8 checkpoint for healthcare research and literature analysis, with 124B total parameters and 5.1B active. It derives from Ling 3.0 Flash Santé BF16. Its FP8 files were published on October 29; the BF16 model had existed since October 19.
01 / SUMMARY
The essential
Maximum input capacity when published by the source.
Documented maximum generation limit.
Verified availability channels.
Primary functional family and verified specialties.
Declared terms for API access, model weights, or self-hosting.
02 / TECHNICAL SHEET
Limits and integration
| API ID | inclusionAI/Ling-3.0-flash-Sante-fp8 |
|---|---|
| Model type | Text and reasoning |
| Access model | Open weights |
| License | MIT |
| Deployment | Self-hosted |
| Release | 29 OCT 2026 |
| Knowledge cutoff | Not published |
| Entrance | Text |
| Exit | Text |
| Context window | 256K tokens |
| maximum output | Not published |
| Reasoning | Enabled by default; configurable |
| Published tools | Features |
| Structured Outings | Not published |
| Batch processing | Not published |
| Prompt cache | Not published |
| Fine-tuning | Not published |
| Verified platforms | Hugging Face · SGLang · vLLM |
Evidence synthesis and document extraction in research settings with professional oversight.
This is a research model, not a medical device. It must not independently diagnose, prescribe or decide treatments. Benchmarks do not establish clinical effectiveness. Protect health data and verify answers with qualified professionals.
03 / CAPABILITIES
What it can do
Orientation
Evidence synthesis and document extraction in research settings with professional oversight.
Context
256K tokens · Not published
Tools and integration
Features
Access
Hugging Face · downloadable checkpoint
Input modalities
Text
04 / PRICES
Documented cost
| Concept | Worth | Unit/condition |
|---|---|---|
| Standard input | No published official API rate | Standard API rate |
| Cached input | Not published | Reading reused prefixes |
| Cache write or storage | Not published | The condition varies by provider |
| Standard output | Self-hosted infrastructure cost | May include reasoning tokens |
| Batch input | Not published | Asynchronous processing |
| Batch output | Not published | Asynchronous processing |
MIT weights for self-hosting, with no verified hosted pricing for this checkpoint. Memory and performance depend on FP8 support and the server; numerical equivalence with BF16 is not assumed.
i Prices change and may depend on level, region or context length. Check the source before making a decision.
05 / EVALUATIONS
How to read the results
“
Provider FP8 results at temperature 1.0, top-p 0.95 and maximum reasoning. BF16 uses the same protocol, but its scores differ. MedMCQA uses the validation set. Inferama has not reproduced these tests.
| Benchmark | Result | Metric | Source |
|---|---|---|---|
| HealthBench Professional · Sante FP8 | 52,9 / 100 | Published score | View source ↗ |
| MedQA-USMLE · Sante FP8 | 94,5 / 100 | Published score | View source ↗ |
Inferama only highlights a “best result” when the metric, test set, configuration, and date allow for an equivalent comparison. The supplier's figures are presented as claims from its own source.
06 / VERSIONS
Chronology
Ling 3.0 Flash Sante FP8
Version added to Inferama's verified catalog.
07 / SOURCES