Common problems & fixes
From the cards (ev-card-32b); this is not an exhaustive list.
- Reasoning text mixed into answers, or tool calls not parsed: start the server with
--reasoning-parser k2_horizon and --tool-call-parser k2_horizon.
- Odd output in plain completion: leave both parsers off.
- Truncated reasoning: allow at least 32,768 output tokens.
- Weaker answers at low effort: set
reasoning_effort="high".
- 375B on SGLang: the validated recipe disables multithreaded weight loading.
- Transformers errors on
generate: drop token_type_ids from the tokenizer output.
- Wrong checkpoint: pass an exact branch name to
--revision (CPT, Cooldown, FTPO, SFT).
Versions, changelog & cadence
Dates come from the hub listing (ev-hub), the JaisVoice README (ev-announce) and the report (ev-report-release).
| Date | Release |
|---|
| 2026-09-29 | Jais-3 weights repositories created (0.9B, 3.7B, 7B, 32B, MoVA-36B-A4B, 375B-A23B) and Uno adapters |
| 2026-09-29 to 2026-10-02 | FP8 and MLX builds |
| 2026-10-06 | Jais-3-Data sample created |
| 2026-10-12 to 2026-10-13 | GGUF builds |
| 2026-10-15 | Release announced |
| 2026-11-06 and 2026-11-12 | Technical report v1 and v2 |
| 2026-11-12 | Weights repositories last modified |
Revisions are Hugging Face commits and stage branches. The repository changelog is the commit history. The cards link no release notes, so the content of the 2026-11-12 update is not documented.
Hugging Face repository discussion tabs (not opened), GitHub Issues on the ifm-ai repositories (ev-code), and the jaischat.ai web service named in the report. Correspondence addresses in the report are at mbzuai.ac.ae. The cards give no support contact.
Known issues
- The cards say the code repository is coming soon, while the report names three public repositories (
ev-card-32b, ev-code).
- The card names (32B, 375B) differ from the safetensors counts (34.8B, 379.2B). The counts include embeddings.
- The cards mark some benchmarks as to be released later.
- The xLLM and RL360 READMEs cover K2 Horizon, so Jais-3 reproduction steps are not documented (
ev-code).
- The K2 Horizon datasets named on the base card return HTTP 401 on the hub API (
ev-k2-card).
The ownership factor this domain covers, drawn from the one entry record.
2
TransparencyDo you know what it is: weights, training, behaviour, and legible terms?
StrongWeights, stage checkpoints, code, evaluation suite, a data sample and a detailed report with named sources are public. The full corpus, partner-supplied books and generator-model outputs are not shared (ev-report-release, ev-report-data, ev-data-sample).
How this scores (AOI sub-dimensions)
Provenance3/5how well we can trace and verify what went into the modelThe weights sit in the verified inception42 organisation as safetensors with LFS sha256 values, a completed hub scan with no flagged files, and a documented lineage from K2 Horizon (ev-hub, ev-report-lineage).
Governance3/5how accountable and well-documented the publisher isMBZUAI, Inception42 and Cerebras are accountable institutions, and the report names its authors and contributors (ev-report-release).
The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.
Model cardread2026-11-19
The 32B card declares license apache-2.0 in its front matter and has no License section.
Model cardread2026-11-19
The 0.9B, 3.7B, 7B, MoVA-36B-A4B and 375B-A23B cards repeat the 32B licence declaration, intended-use text and limitations.
Licenceread2026-11-19
Apache-2.0 is declared in the front matter of all six weights cards and in section 7.1 of the technical report.
Technical_reportread2026-11-19
The report says each Jais 3 model starts from the final post-trained K2 Horizon checkpoint of the same size, keeps its architecture and tokenizer, and is adapted by continual pretraining, SFT, FTPO and RLVR.
Technical_reportread2026-11-19
The report lists released weights, stage checkpoints, training and RL code, a training-data sample, run configurations, W&B logs, audited benchmarks and evaluation code (Table 41).
Technical_reportread2026-11-19
The report names each source with token counts: ArabicTxT360 web text (28.4B), literature collections (9.65B), translated reasoning (30.9B), parallel dialects (26.9B), Arabizi (25.1B) and 11.9B tokens of partner-supplied Arabic books with no named source.
Technical_reportread2026-11-19
The report says GLM-5, Jais 2, Qwen3.8-27B, Qwen3.6-27B, gemma-4-31B-it and MiniMax-M3 generated or judged parts of the training data.
Technical_reportread2026-11-19
Section 6.5 reports publisher-run results on AraSafe (1,254 harmful prompts, above 92% refusal for every size), Arabic Regional Safety, HarmBench and XSTest, scored with gemma-4-26B-A4B-it.
Technical_reportread2026-11-19
The report says Jais 3 leads or matches open models of its size on Arabic benchmarks.
Technical_reportread2026-11-19
Training covers Gulf, Egyptian, Levantine and Darija Arabizi, country knowledge for 22 states, and translation for seven dialects.
Documentationread2026-11-19
All six weights repositories are ungated, hold safetensors only with LFS sha256 values, and report a completed hub scan with no flagged files.
Documentationread2026-11-19
Each weights repository has branches CPT, Cooldown, FTPO and SFT besides main.
Documentationread2026-11-19
The IFM organisation hosts FP8, MLX and GGUF builds of the Jais-3 models, each tagged Apache-2.0.
Documentationread2026-11-19
IFM/Jais-3-Data is an ungated 23.8B-token sample, about 10% of CPT and SFT tokens, in four Arabic subsets licensed ODC-BY or CC BY 4.0.
Documentationread2026-11-19
The xllm, RL360 and gen-eval-arabic repositories are public with Apache-2.0 LICENSE files.
Documentationread2026-11-19
The JaisVoice README dates the Jais 3 release to 2026-10-15.
Model cardread2026-11-19
The K2-Horizon-32B card declares Apache-2.0, says "Fully open" and links code and data.
Technical_reportread2026-11-19
The K2-Horizon abstract describes six fully open models, about 10T synthetic pre-training tokens and a release that follows the LLM360 definition of open source (publisher claims).
Licenceread2026-11-19
The MiniMax Community License grants use for non-commercial purposes and sets conditions on commercial use of the software and its derivative works.
Model cardread2026-11-19
The four Jais-3 Uno adapter repositories carry no licence tag and no licence statement.
Documentationunverified2026-11-19
The ifm.ai blog and tech-report pages return HTTP 403 (Cloudflare challenge), so their text is unavailable.