rekaai
Reka Flash 3
rekaai/reka-flash-3
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Reka Flash 3 (rekaai/reka-flash-3) is a 21-billion-parameter, text-only language model from Reka, released in a Llama-compatible format under an Apache-2.0 license. According to its model card, it was trained from scratch with supervised finetuning followed by RLOO, and it produces a reasoning trace before answering; the OpenRouter catalog lists reasoning as mandatory, a 65,536-token context window, and function-calling support. The vendor describes it as competitive for its size, but this source pack contains no independent evaluations.[1][2]
Strengths
- Catalog-described strengths include general chat, coding tasks, instruction-following, and function calling; structured outputs and reasoning controls are among its supported parameters.[1]
- Trained from scratch using supervised finetuning plus RLOO with model-based and rule-based rewards; emits a tagged reasoning trace whose length can be capped via budget forcing.[2]
- Distributed in a Llama-compatible format that runs with standard tooling such as Hugging Face transformers or vLLM; the vendor positions it for low-latency or on-device deployments.[2]
Limitations
- The vendor advises treating it as English-only: it often reasons in English for non-English prompts, which can degrade output quality in other languages.[2]
- The model card states it is not well suited to knowledge-intensive tasks and recommends pairing it with web search; it has not undergone extensive alignment or persona training.[2]
- Evidence here is limited to the vendor's model card and the OpenRouter catalog; performance claims are vendor-reported and unverified, so validate on your own workloads.[1][2]
Where to use it
- English-language assistants or agents where an explicit reasoning step helps and a tunable thinking budget is useful.[2]
- Self-hosted or latency-sensitive deployments using Llama-compatible tooling, matching the vendor's stated positioning.[2]
- Chat, coding help, and function-calling workflows, as suggested by the catalog description; pair with web search for knowledge-heavy tasks.[1][2]
Sources
- [1] OpenRouter catalog · Reka Flash 3Sources checked: Nov 13, 2026
- [2] Model documentation · Reka Flash 3Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Reka Flash 3 is a general-purpose, instruction-tuned large language model with 21 billion parameters, developed by Reka. It excels at general chat, coding tasks, instruction-following, and function calling. Featuring a...
Specifications
- Model ID
- rekaai/reka-flash-3
- Creator
- rekaai
- Added to OpenRouter
- 12.03.2025
- Context length
- 65,536
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.100 / out $0.200
- Knowledge cutoff
- 31.01.2025
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
No evaluations are supplied for this model.
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| Reka | 65,536 | $0.100 | $0.200 | — | — | — |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.