Up to date
108B parameters (8 of 128 experts per token), images and text in, text out, 64K tokens of context. 2 providers serve it from $0.60 per million input tokens.
Verified 2026-10-25 Rank #156 in LLMs · score 91.9 method v1 Sources 4 Downloads 30d 25.2k · #440 of 619 tracked
Picked by rule from 8 tracked pieces: the lab's own post, the most shared, a field report, the latest.
The GLM-4.5V model requires removing specific tags from its fine-tuning dataset.
ZhipuAI released the GLM-4.6V series, featuring a 106B foundation model and a 9B Flash variant optimized for local deployment.
GLM-4.5V is Zhipu AI's August 2025 open-weight vision-language MoE with 106B total parameters and 12B activated per token.
The 3 picks above, plus these 5. 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.
2 providers with a live price, from $0.60 to $0.60 per million input tokens; the median is $0.60. One dot per provider, cheapest in teal.
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 2 API providers; 10 quantized builds for local use.
Top Spaces running it: jairwaal/image ♥ 16 · umint/openwebui ♥ 6 · akhaliq/zai-org-GLM-4.5V ♥ 5 · Jakob08/moneychatbot ♥ 3
Weights only, an estimate: leave room for the KV cache. Method
About 65 GB at INT4, 129 GB at FP8, 259 GB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2025-08-10), amber triangles are 5 dated articles (each links to it), grey ticks are repo commits.
15 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.
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 | image-text-to-text |
|---|---|
| Library | transformers |
| Architectures | Glm4vMoeForConditionalGeneration |
| Model type | glm4v_moe |
| Dtype | bfloat16 |
| Params by dtype | BF16: 107,710,927,360 · F32: 5,760 |
| Size band | XXL |
| MoE | yes |
| Experts | 128 |
| Experts per token | 8 |
| Context length | 65,536 (config.json max_position_embeddings) |
| Languages | zh, en |
| Papers | arXiv:2507.01006 |
| License | MIT (mit) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · Glm4vMoeForConditionalGeneration · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Glm4vMoeForConditionalGeneration · 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 GLM-4.5V: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 259 GB · Multi-GPU node (<640 GB) |
|---|---|
| FP8 (1 byte/param) | 129 GB · Multi-GPU node (<640 GB) |
| INT4 (0.5 bytes/param) | 65 GB · Single high-end GPU (<80 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra. MoE: all experts must be resident, so total parameters count.
Base model: zai-org/GLM-4.5-Air-Base
quantized: 10finetune: 5
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| mradermacher/GLM-4.5V-i1-GGUF | quantized | GGUF | — | — | mit | mradermacher | 1 | 231.8k |
| QuantTrio/GLM-4.5V-AWQ | quantized | AWQ | 16.9B | L | mit | QuantTrio | 19 | 23.3k |
| mradermacher/GLM-4.5V-GGUF | quantized | GGUF | — | — | mit | mradermacher | 1 | 8.4k |
| tiny-random/glm-4v-moe | finetune | — | 5.85M | S | — | tiny-random | 2 | 4.3k |
| mlx-community/Huihui-GLM-4.5V-abliterated-mxfp4 | quantized | mxfp4 | 108B | XXL | mit | mlx-community | 3 | 2.9k |
| ggml-org/GLM-4.5V-GGUF | quantized | GGUF | — | — | mit | ggml-org | 5 | 2.5k |
| INC4AI/GLM-4.5V-int4-AutoRound | quantized | int4 | 2.12B | M | — | INC4AI | 14 | 1k |
| mlx-community/GLM-4.5V-abliterated-8bit | quantized | -8bit | — | — | mit | mlx-community | 1 | 1.5k |
| mlx-community/GLM-4.5V-abliterated-4bit | quantized | -4bit | — | — | mit | mlx-community | 1 | 1.2k |
| mlx-community/Huihui-GLM-4.5V-abliterated-5bit | quantized | — | — | — | mit | mlx-community | 0 | 1.1k |
| yujiepan/glm-4.5v-tiny-random | finetune | — | 5.85M | S | — | yujiepan | 0 | 875 |
| yujiepan/glm-4v-moe-tiny-random | finetune | — | 5.85M | S | — | yujiepan | 0 | 857 |
| turboderp/GLM-4.5V-exl3 | quantized | exl3 | — | — | mit | turboderp | 5 | 220 |
| zeromodels/glm-4.5v | finetune | — | — | — | mit | zeromodels | 0 | 459 |
| tiny-random/glm-4.5v | finetune | — | 5.85M | S | — | tiny-random | 0 | 118 |
| Date | Commit | SHA |
|---|---|---|
| 2025-10-25 | update for new format config of transformers lib / sglang lib | ed47433 |
| 2025-10-25 | Merge remote-tracking branch 'origin/main' | 9081159 |
| 2025-10-25 | update for new format config of transformers lib / sglang lib | f33f585 |
| 2025-09-23 | Update tokenizer_config.json | 5b59df8 |
| 2025-08-18 | update | a78e5f2 |
| 2025-08-15 | Improve model card: add detailed overview, expanded usage, and project updates (#7) | 2fb7768 |
| 2025-08-12 | paper | 8a49973 |
| 2025-08-11 | update readme | dcfb7bf |
| 2025-08-11 | update | 2a76d0e |
| 2025-08-10 | Add files using upload-large-folder tool | 8a552c2 |
| 2025-08-10 | Add files using upload-large-folder tool | ce58144 |
| 2025-08-10 | Add files using upload-large-folder tool | ee8faeb |
| 2025-08-10 | initial commit | 0974377 |
Hosted endpoints observed for GLM-4.5V, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting. Cheapest observed input price: $0.60/M.
| Provider | Model ref | Context | Quant | $/M in | $/M out | tok/s | TTFT | Tools | Status | Observed |
|---|---|---|---|---|---|---|---|---|---|---|
| Novita hf, openrouter | zai-org/glm-4.5v | 64K | fp8 | $0.60 | $1.80 | 68 | 1258 ms | yes | live | 2026-10-25 |
| Z.ai hf, openrouter | glm-4.5v | 64K | fp8 | $0.60 | $1.80 | 50.1 | 2142 ms | yes | 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 21:35 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:03 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.
