Up to date
HuggingFaceH4 MIT out 3 years ago
7.24B parameters, text in, text out, 32K tokens of context. 1 provider serves it.
Verified 2026-10-25 Rank #186 in LLMs · score 89.8 method v1 Sources 3 Downloads 30d 98.7k · #317 of 619 tracked
Changed this week: 3 new quantized builds on the Hub (37 in all).2026-10-24
Picked by rule from 12 tracked pieces: the lab's own post, the most shared, a field report, the latest.
Golioth enables zero-install embedded training for Zephyr using Kasm and Docker to normalize development environments.
The 4 picks above, plus these 8. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.
| # | Cheapest live row | In /1M | Out /1M | Context | Quant | Seen |
|---|---|---|---|---|---|---|
| 1 | Featherless | — | — | — | — | Oct 25 |
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
Served by 1 API provider; 37 quantized builds for local use; a big base for builders (658 fine-tunes, adapters and merges).
Top Spaces running it: HuggingFaceH4/zephyr-chat ♥ 903 · fffiloni/consistent-character ♥ 646 · fffiloni/image-to-music-v2 ♥ 578 · allenai/reward-bench ♥ 438
Weights only, an estimate: leave room for the KV cache. Method
About 4.3 GB at INT4, 8.7 GB at FP8, 17 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2023-10-26), amber triangles are 5 dated articles (each links to it), grey ticks are repo commits.
695 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| author-reported AlpacaEval · Chat and follow instructions | 90.6 percent | Zephyr-7b-β | model author | 2026-10-07 |
| author-reported MT-Bench · Chat and follow instructions | 7.34 | Zephyr-7b-β | model author | 2026-10-07 |
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in LLMs
| Pipeline tag | text-generation |
|---|---|
| Library | transformers |
| Architectures | MistralForCausalLM |
| Model type | mistral |
| Dtype | bfloat16 |
| Params by dtype | BF16: 7,241,732,096 |
| Size band | M |
| MoE | no |
| Context length | 32,768 (config.json max_position_embeddings) |
| Languages | en |
| Papers | arXiv:2305.18290, arXiv:2310.16944, arXiv:2305.14233, arXiv:2310.01377 |
| License | MIT (mit) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · MistralForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · MistralForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Weights-only memory for zephyr-7b-beta: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 17 GB · Single GPU (<24 GB) |
|---|---|
| FP8 (1 byte/param) | 8.7 GB · Single GPU (<24 GB) |
| INT4 (0.5 bytes/param) | 4.3 GB · Single GPU (<24 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
Base model: mistralai/Mistral-7B-v0.1
quantized: 37merge: 89finetune: 147adapter: 422
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| TheBloke/zephyr-7B-beta-AWQ | quantized | AWQ | 7.24B | M | mit | TheBloke | 39 | 1.2M |
| TheBloke/zephyr-7B-beta-GGUF | quantized | GGUF | — | — | mit | TheBloke | 230 | 342.6k |
| TheBloke/zephyr-7B-beta-GPTQ | quantized | GPTQ | 7.24B | M | mit | TheBloke | 58 | 193.6k |
| RedHatAI/zephyr-7b-beta-marlin | quantized | — | 1.19B | S | — | RedHatAI | 0 | 52.9k |
| Kabster/BioMistral-Zephyr-Beta-SLERP | merge | — | 7.24B | M | apache-2.0 | Kabster | 2 | 47.4k |
| ghost-x/ghost-7b-v0.9.0 | finetune | — | 7.24B | M | mit | ghost-x | 5 | 38.1k |
| ContinuousAT/Zephyr-CAT | adapter | — | — | — | — | ContinuousAT | 1 | 29.8k |
| EmbeddedLLM/Mistral-7B-Merge-14-v0.3 | merge | — | 7.24B | M | apache-2.0 | EmbeddedLLM | 8 | 19.2k |
| EmbeddedLLM/Mistral-7B-Merge-14-v0.3-ft-step-15936 | merge | — | 7.24B | M | apache-2.0 | EmbeddedLLM | 0 | 17k |
| EmbeddedLLM/Mistral-7B-Merge-14-v0.3-ft-step-9984 | merge | — | 7.24B | M | apache-2.0 | EmbeddedLLM | 0 | 17k |
| MexIvanov/zephyr-python-ru-merged | finetune | — | 7.24B | M | mit | MexIvanov | 2 | 12.9k |
| jambroz/sixtyoneeighty-7b | merge | — | 7.24B | M | apache-2.0 | jambroz | 2 | 12k |
| second-state/Zephyr-7B-Beta-GGUF | quantized | GGUF | — | — | mit | second-state | 0 | 11.7k |
| BioMistral/BioMistral-7B-Zephyr-Beta-SLERP-GGUF | merge | GGUF | — | — | — | BioMistral | 3 | 9.8k |
| mradermacher/zephyr-7b-beta-GGUF | quantized | GGUF | — | — | mit | mradermacher | 0 | 9.5k |
| MaziyarPanahi/zephyr-7b-beta-GGUF | quantized | GGUF | — | — | apache-2.0 | MaziyarPanahi | 0 | 8.1k |
| QuantFactory/zephyr-7b-beta-GGUF | quantized | GGUF | — | — | mit | QuantFactory | 0 | 7.8k |
| grimjim/zephyr-beta-wizardLM-2-merge-7B | merge | — | 7.24B | M | apache-2.0 | grimjim | 1 | 5.4k |
| grimjim/fireblossom-32K-7B | merge | — | 7.24B | M | cc-by-nc-4.0 | grimjim | 3 | 5.2k |
| kukidevalml/LORAs | adapter | GGUF | 1.26B | S | apache-2.0 | kukidevalml | 2 | 4.1k |
| DZgas/GIGABATEMAN-7B | merge | — | 7.24B | M | — | DZgas | 8 | 3.2k |
| DevQuasar-8/HuggingFaceH4.zephyr-7b-beta-GGUF | quantized | GGUF | — | — | — | DevQuasar-8 | 0 | 3.3k |
| tensorblock/zephyr-7b-beta-GGUF | quantized | GGUF | — | — | mit | tensorblock | 0 | 3.1k |
| OPTML-Group/SimNPO-WMDP-zephyr-7b-beta | finetune | — | 7.24B | M | mit | OPTML-Group | 0 | 2.8k |
| CreitinGameplays/ConvAI-9b | finetune | — | 8.99B | M | mit | CreitinGameplays | 1 | 2.4k |
| papahawk/devi-7b | finetune | — | 7.24B | M | mit | papahawk | 0 | 2k |
| OPTML-Group/NPO-WMDP | finetune | — | 7.24B | M | mit | OPTML-Group | 0 | 1.8k |
| yatinece/yk_8x7b_model_v1 | merge | — | 46.7B | XL | apache-2.0 | yatinece | 0 | 1.8k |
| yweslakarep/huh-1 | merge | — | 7.24B | M | apache-2.0 | yweslakarep | 1 | 1.7k |
| azale-ai/Starstreak-7b-beta | finetune | — | 7.24B | M | cc0-1.0 | azale-ai | 5 | 1.4k |
| BioMistral/BioMistral-7B-Zephyr-Beta-SLERP | merge | — | 7.24B | M | — | BioMistral | 2 | 1.5k |
| OPTML-Group/GradDiff-WMDP | finetune | — | 7.24B | M | mit | OPTML-Group | 0 | 1.5k |
| OPTML-Group/NPO-SAM-WMDP | finetune | — | 7.24B | M | mit | OPTML-Group | 0 | 1.4k |
| arvnoodle/hcl-zephyr-7b-javascript-lotuscript-GGUF | quantized | GGUF | — | — | apache-2.0 | arvnoodle | 0 | 1.3k |
| richardyoung/zephyr-7b-beta-abliterated | finetune | — | 7.24B | M | mit | richardyoung | 1 | 931 |
| arcee-ai/BioMistral-merged-zephyr | merge | — | 7.24B | M | — | arcee-ai | 1 | 881 |
| MaziyarPanahi/zephyr-7b-beta-GPTQ | finetune | GPTQ | 1.20B | S | apache-2.0 | MaziyarPanahi | 1 | 873 |
| lvcalucioli/zephyr-7b-beta_multiple-choice | adapter | — | — | — | mit | lvcalucioli | 0 | 848 |
| AbteeXAILab/lumynax-infused-zephyr-7b-beta-gguf | quantized | gguf | — | — | mit | AbteeXAILab | 0 | 839 |
| DhruvalLabs/zephyr-7b-beta-GGUF | quantized | GGUF | — | — | mit | DhruvalLabs | 1 | 745 |
| Date | Commit | SHA |
|---|---|---|
| 2024-10-16 | Add UltraChat and UltraFeedback references | 892b3d7 |
| 2024-09-23 | Delete training_args.bin | 20e1a58 |
| 2024-02-29 | Update widgetData input (#54) | b70e0c9 |
| 2023-12-14 | Update README.md | dc24cab |
| 2023-12-14 | Update Open LLM Leaderboard source urls (#42) | f4e901c |
| 2023-12-01 | [Feedback welcome] Add evaluation results to model card metadata (#40) | 0f17b36 |
| 2023-11-22 | Add widget example (#34) | 8d44a71 |
| 2023-11-21 | Update README.md | 36db76e |
| 2023-11-20 | Adding Evaluation Results (#31) | c2339b6 |
| 2023-11-20 | Update README.md (#28) | abe3655 |
| 2023-11-04 | Upload MistralForCausalLM (#20) | 8af01af |
| 2023-10-27 | Update README.md (#1) | 3bac358 |
| 2023-10-27 | Update README.md | af28f1d |
| 2023-10-27 | Update README.md | 99dc9c4 |
| 2023-10-26 | Update README.md | 8d4f1f7 |
| 2023-10-26 | Update README.md | 0f4b871 |
| 2023-10-26 | Update README.md | e981ffb |
| 2023-10-26 | Update config.json | e323593 |
| 2023-10-26 | Delete dialogue_template.json | fff0b6c |
| 2023-10-26 | Update README.md | 63bc8ee |
| 2023-10-26 | Update README.md | 8bf7bff |
| 2023-10-26 | Duplicate from HuggingFaceH4-colab/zephyr-7b-beta-sharded | 97ecdb4 |
Hosted endpoints observed for zephyr-7b-beta, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting.
| Provider | Model ref | Context | Quant | $/M in | $/M out | tok/s | TTFT | Tools | Status | Observed |
|---|---|---|---|---|---|---|---|---|---|---|
| Featherless hf | HuggingFaceH4/zephyr-7b-beta | — | — | — | — | — | — | — | live | 2026-10-25 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:36 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:04 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
