Alex Getmansegnali precoci sull'IA

GLM-5

Z.ai

Rilasciato Aggiornato
Finestra di contesto
203K
202.752 token
Output massimo
131K
131.072 token
Ingresso / Uscita
testo → testo

Prezzi API

USD / 1 milione di token

IngressoUscitaIngresso in cache
  • $1,00
zai-pricing
  • $3,20
zai-pricing
  • $0,20
zai-pricing

Cronologia

  1. · Versione dei pesi · link di riscontro
Prezzo ingresso/uscita · USD / 1 milione di token
ModelloLink di riscontroPrezzo ingresso/uscita
glm-5
standard
zai-pricing
Percorso API non disponibile
$1,00 / $3,20
  1. · Link di riscontro · GitHub search · Z.ai · pulls · smg-project/smg · @hello-alexmcc
    Estratto
    Atria-Dawn, `glm`, beside Atria-Dawn-Preview: the template writes the GLM-5 shape (`[gMASK]<sop>`, a reasoning-effort system turn, `<|system|>` / `<|user|>` / `<|assistant|>` / `<|observation|>` turns), the tools
  2. · Link di riscontro · GitHub search · Z.ai · pulls · milabench/milabench · @bouthilx
    Estratto
    - GLM-5: layout check before launch, loss from the last stage, checkpoint timing and checksums, crash cleanup; `benchmarks/megatron/dev.yaml`, `scripts/trillium/glm/`.
  3. · Link di riscontro · GitHub search · Z.ai · pulls · maximhq/bifrost · @mickgvirtu
    Estratto
    is not `schemas.Anthropic`, and the served model -- `glm-5`, `Kimi-K2`, ... -- is neither opus-4.8 nor Fable), so Claude Code's per-turn reminders were hoisted every
  4. · Link di riscontro · GitHub search · Z.ai · pulls · anomalyco/models.dev · @models-dev-automation[bot]
  5. · Link di riscontro · GitHub search · Z.ai · pulls · ai-dynamo/frontend-crates · @dundysm
  6. · Link di riscontro · GitHub search · Z.ai · pulls · ai-dynamo/dynamo · @indrajit96
    Estratto
    for any DSA model (including DeepSeek V3.2 and GLM-5), or matching `flashmla_kv`/`trtllm` kernels on K3.1's sharded DCP path. Kernels SGLang has not resolved at CLI
  7. · Link di riscontro · GitHub search · Z.ai · pulls · ROCm/ATOM · @whx-sjtu
  8. · Link di riscontro · GitHub search · Z.ai · pulls · NVIDIA/TensorRT-LLM · @moraxu
    Estratto
    class is registered for both `DeepseekV3ForCausalLM` and `DeepseekV32ForCausalLM`; GLM-5 (`GlmMoeDsaForCausalLM`) and Mistral Large 3 subclass it, so exact-class matching excludes them. Kimi K2 declares `DeepseekV3ForCausalLM`
  9. · Link di riscontro · GitHub search · Z.ai · pulls · NVIDIA/TensorRT-LLM · @kaiyux
    Estratto
    - `modeling_glm5_next_vision.py`: the GLM-5 input processor takes the prefix tokenization cache (the fourth commit). Before, it tokenized every text prompt in full and ignored `enable_tokenization_cache`.
Specifiche

GLM-5 è un modello di Z.ai con pesi aperti, rilasciato il 2026-02-12. La finestra di contesto è di 202.752 token. L’output massimo è di 131.072 token. Il prezzo per milione di token è $1,00 per l’input e $3,20 per l’output.

Finestra di contesto
202.752
Output massimo
131.072
Ingresso
testo
Uscita
testo
Pesi
Aperti
Ingresso · USD / 1 milione di token
$1,00
Uscita · USD / 1 milione di token
$3,20
Ingresso in cache · USD / 1 milione di token
$0,20

Domande

Qual è la finestra di contesto di GLM-5?

La finestra di contesto è di 202.752 token. L’output massimo è di 131.072 token.

Quanto costa GLM-5?

Il prezzo per milione di token è $1,00 per l’input e $3,20 per l’output. L'ingresso in cache costa $0,20 per milione.

GLM-5 ha pesi aperti?

Sì. Z.ai pubblica i pesi.

Quando è stato rilasciato GLM-5?

Il 2026-02-12.