GLM-5.3 é Lançado: O Vazamento Era Real — O Carro-Chefe de Codificação e Defesa Cibernética Pós-Treinado da Z.ai

The leak was real, the launch is here, and GLM-5.3 now has an independent score to argue about. Artificial Analysis' Intelligence Index — measured by the lab, not by Z.ai — puts GLM-5.3 at 60, tied with Kimi K3 for the top open-weights score on the board and 7 points clear of GLM-5.2's 53. The API went live this week at the same price as its predecessor, the open weights are confirmed for Friday, August 28, and the coding and cyber-defense claims Z.ai has been making since the August 14 announcement are starting to become testable. This page first tracked GLM-5.3 from its August 3 leak traces; this is the launch report, updated in place with what the launch, the API, and the first independent benchmark actually confirmed.

Do vazamento ao lançamento

The four traces that surfaced on August 3 — a "ZCode for GLM-5.3" harness page, an official docs page reachable for roughly an hour, a Bing index entry reading "GLM-5.3 Official Harness," and a commit adding a "glm-5.3" entry with JSON Schema support to Zhipu's official Java SDK — all pointed at a real, named release. Z.ai co-founder Tang Jie's "sooooooon" reply and the "epic-level plus" framing are now confirmed by an actual product rather than a rumor. The "roughly a week" timing signal this page tested — posted on X by @teortaxesTex after DeepSeek V4 Pro shipped on August 13 — held to the day: Z.ai formally announced GLM-5.3 on August 14 under the slogan "Built to Code. Ready for Cyber Defense." The follow-on came this week: on August 19 Z.ai said the GLM-5.3 API was live and open for calls, priced the same as GLM-5.2, with the model already wired into ZCode, AutoClaw, and the GLM Coding Plan.

O que o pós-treinamento realmente comprou

A alegação arquitetural é a parte que vale a pena fixar, porque {{1}}o detalhe mais ruidoso do vazamento — parâmetros ultrapassando um trilhão — está errado{{/1}}. O GLM-5.3 não é um modelo maior. A Z.ai afirma que ele reutiliza exatamente a mesma base de Mixture-of-Experts de 743B do GLM-5.2 (aproximadamente 40 bilhões de parâmetros ativos por token), mantém a mesma janela de contexto de 1M de tokens e aproximadamente 128K de saída máxima, e obtém todos os seus ganhos a partir de pós-treinamento em escala ampliada: mais ambientes de tarefas de horizonte longo, mais tipos de ambiente e execuções de treinamento mais longas, construídos sobre o IndexShare de contexto longo, o SAO de RL assíncrono e o framework de código aberto slime que já produziu o GLM-5.2. Esse enquadramento de "sem retreinamento, apenas pós-treinamento" é da própria Z.ai, e não foi auditado de forma independente. O tamanho é o outro destaque: com 743B de parâmetros totais e apenas cerca de 40B ativos por token, o GLM-5.3 é leve o suficiente para ser auto-hospedado em um cluster modesto e barato o suficiente para ser servido em volume — o posicionamento de "menor, mais barato, aberto" que os comentários do dia do lançamento atribuíram a ele, e um contraste nítido com os carros-chefe fechados de fronteira com os quais está sendo comparado.

Codificação e agentes: os números que a Z.ai alega

Em codificação, a Z.ai reporta — tudo relatado pelo fornecedor e não reproduzido — Terminal-Bench 3.0 subindo de 4.6 para 28.3, que a Z.ai chama de melhor pontuação de pesos abertos nesse harness; DeepSWE v1.1 subindo de 46.2 para 66.9; SWE-Marathon aproximadamente dobrando de 19.4 para 42.5; e Agents' Last Exam (CLI) subindo de 23.8 para 28.5. No bench de código interno da Z.ai, o GLM-5.3 pontuou 31.4% com cerca de 50K tokens de saída por tarefa em alto esforço, contra 29.5% do Claude Opus 4.8 com cerca de 120K tokens — o argumento da Z.ai é que o GLM-5.3 alcança um resultado comparável gastando muito menos tokens de saída. O Claude Fable 5 ainda lidera essa avaliação interna com 39.5% em esforço máximo, e a Z.ai admite que o GLM-5.3 ainda fica atrás do GPT-5.6 Sol e do Claude Fable 5 em várias avaliações de codificação mais difíceis. Trate todos esses números como dados do fornecedor até que um harness independente os reproduza.

Defesa cibernética: a capacidade que ninguém viu chegar

Os números de segurança cibernética são a verdadeira notícia, e também são inteiramente relatados pelos fornecedores. No CyberGym, um benchmark de descoberta e validação de vulnerabilidades em caixa-branca, a Z.ai relata GLM-5.3 com 84,5%, acima dos 77,2% do GLM-5.2 e à frente do Mythos 5, da Anthropic (83,8%) e do GPT-5.6 Sol (83,6%). No ExploitBench, que exige tanto análise de causa raiz quanto um exploit funcional, o GLM-5.3 mais que dobrou de 24,4% para 54,4%, embora o Mythos 5 (78,0%) continue à frente. No ExploitGym, a Z.ai relata 105 tarefas concluídas em um orçamento de 2 horas e 130 em 6 horas, contra 29 e 39 para o GLM-5.2 — novamente atrás do Mythos 5 (181 e 247). A Z.ai enquadra a capacidade cibernética como uma propriedade emergente do pós-treinamento em escala — "a capacidade continuou a se acumular à medida que o treinamento escalava", nas palavras da empresa — em vez de um alvo deliberado.

A Z.ai adiciona uma alegação de mundo real para acompanhar os benchmarks: em testes com equipes de segurança, o GLM-5.3 identificou 2.436 vulnerabilidades em 269 projetos open-source, 1.097 delas classificadas como críticas ou de alta gravidade, com a descoberta mais antiga datando de 1981 e uma "vida útil" média de 26,6 anos. Esse é o número mais impressionante do anúncio e também o menos verificável de forma independente. A empresa combinou a capacidade com um Security Disclosure Ledger para divulgação coordenada, um programa de "acesso confiável" que limita funções cibernéticas sensíveis a usuários verificados e uma iniciativa "Open Source Shield" para auditar continuamente os principais projetos open-source.

A baseline GLM-5.3 tinha que superar

GLM-5.2 is the reference point the whole story hangs on. It shipped in June 2026 as a 743B Mixture-of-Experts model with roughly 40B active parameters per token, a 1M-token context window, a 128K max output, an MIT license, and open weights on Hugging Face. Independently, Artificial Analysis' Intelligence Index puts GLM-5.2 at 53 — the highest open-weights score on the index until this week. GLM-5.3 now clears it by 7 points: Artificial Analysis measures GLM-5.3 at 60 on the same index (v4.1.1), tying Kimi K3 for the top open-weights position and landing it in the frontier band alongside closed flagships like Claude Fable 5 and GPT-5.6 Sol. That is the first independent number attached to GLM-5.3, and it is consistent with the direction — if not every detail — of Z.ai's own claims. On long-horizon coding, the OrcaRouter harness measures 77.9 on Terminal-Bench 2.1, while Z.ai's best-reported GLM-5.2 figure is 82.7, which would be the first open-weight score above 80 but is vendor-reported and unreproduced. The list price is $1.40 per million input and $4.40 per million output tokens.

The scoreboard above is the leak-era projection this page published before launch — the ">1T params (rumored)" row, the unconfirmed context and license, the projected AA index. The launch corrected the biggest cell: GLM-5.3 reuses the same 743B base as GLM-5.2, so there is no parameter jump. The context window is confirmed at 1M, and the license stays unconfirmed because the open weights have not shipped yet. The projected index cell — this page's own guess of ~57–60 — was the rare projection that came in on the nose: the real number is 60, and the open question now is what happens when that score is reproduced against the actual weights.

The capture above is the independent baseline GLM-5.3's claims are measured against. GLM-5.2 tops the open-weights leaderboard at an Artificial Analysis Intelligence Index of 53. The first test of whether GLM-5.3's post-training deltas move that number has now arrived: Artificial Analysis measures GLM-5.3 at 60 on the same index — tied with Kimi K3 for the open-weights lead, 7 points ahead of GLM-5.2, and reported by the lab as independently measured.

Preços e disponibilidade

GLM-5.3 is priced identically to GLM-5.2: $1.40 per million input and $4.40 per million output tokens (¥8 / ¥28 in the domestic listing), with cached-input reads at $0.26 / ¥2 per million. Z.ai announced the API was open on August 19, and it is reachable through Z.ai's own API, ZCode, AutoClaw, the GLM Coding Plan, and several partner gateways. One behavior change matters for API callers: requests now require "thinking" enabled across three effort levels — low, high, and max — with no off switch, a breaking change for existing integrations.

Same price does not mean same bill. GLM-5.3 runs roughly 20% more tokens per task than GLM-5.2 did on the same workloads, which a cost-per-task reading puts at about $0.68 against GLM-5.2's $0.44 — still under Kimi K3 (about $0.84) and GPT-5.6 Sol (about $1.23). That per-task math is a derived estimate from observed token usage, not a vendor figure, but it is the number that decides whether the flat $1.40 / $4.40 rate card actually saves you money.

O que o lançamento muda para você

For API callers already on GLM-5.2, the practical step is a model-name change, not a project: GLM-5.2 is OpenAI-compatible and the integration carries over, with the thinking-effort caveat above. For self-hosters, the timeline is now a date rather than a guess: Zhipu promised the weights "two weeks after release" on August 14, which lands on Friday, August 28, and the open question is whether the license stays permissive. For anyone comparing models in the DeepSeek V4 Pro, Qwen3.8-Max, Kimi K3, GPT-5.6 Sol, and Claude Fable 5 tier, GLM-5.3 is now a live, independently scored variable in that ranking instead of a rumor.

The launch-day argument around GLM-5.3 is that the coding frontier has converged: for most everyday tasks, the story goes, few users can reliably tell GPT-5.6 Sol, Claude Fable 5, Kimi K3, GLM-5.2, and Qwen3.8-Max apart. If that convergence is real, the deciding factors stop being raw capability and become price, openness, and switching cost — which is exactly the corner GLM-5.3 is staking out at $1.40 / $4.40 per million on a self-hostable 743B base with weights confirmed for August 28. Whether coding models are genuinely interchangeable is an opinion, not a benchmark; the prices, the parameter count, and the weight date are not.

On the routing side, GLM-5.3 went live on OrcaRouter on August 18, the same day Z.ai's API opened — at the first-party list price, $1.40 / $4.40 per million, passed through with zero markup. The screenshot below shows GLM-5.2's page, which is exactly the shape GLM-5.3 now has: same price, same 1M-token context, same 128K max output. Routing a slice of real traffic to GLM-5.3 with automatic failover to GLM-5.2 or another proven model is a configuration change, not a rewrite — same key, no second contract. If the new model regresses on your workload, the router falls back before a page turns, and you get a quality signal on your own traffic instead of a vendor's slide. For a model whose flagship claims are still mostly vendor-reported, that is the low-risk way to find out for yourself.

O que assistir em seguida

• The weights, on Friday, August 28, and the license line on the model card — permissive MIT like GLM-5.2, or something narrower. Zhipu's cyber-safety hardening is the stated reason for the two-week delay, and the "trusted access" program suggests some functions will be gated regardless.

• Whether the cyber claims hold up outside Z.ai's own harness. The 2,436-vulnerability real-world claim and the CyberGym lead are the numbers independent labs will probe first; the AA Intelligence Index measures general capability, not security.

• Where the index lands once the weights are out. The 60 is scored against the served API; the self-hosted version, with a license attached, is the one teams will actually redeploy.

• DeepSeek V4 Flash's announced price increase, which sets the pricing envelope GLM-5.3 is being judged against, and GPT-5.6 Sol's one-point lead at 61.

Perguntas Frequentes

O GLM-5.3 é um modelo maior que o GLM-5.2?

Não. A Z.ai diz que a GLM-5.3 usa a mesma base Mixture-of-Experts de 743 bilhões de parâmetros que a GLM-5.2, com a mesma janela de contexto de 1 milhão de tokens e aproximadamente 40 bilhões de parâmetros ativos por token. Todos os ganhos relatados vêm de pós-treinamento em escala, não de um aumento de parâmetros — o que corrige diretamente o boato da era dos vazamentos de uma base acima de um trilhão. Essa alegação de arquitetura é da própria Z.ai e não foi auditada de forma independente.

Quando os pesos abertos do GLM-5.3 estarão disponíveis?

Friday, August 28. Zhipu promised the weights "two weeks after release" when it announced GLM-5.3 on August 14, and said "next Friday" when the API went live on August 19 — both readings land on the same date. The license has not been confirmed, and Z.ai has said sensitive cyber functions will be restricted to a verified-user "trusted access" program.

Como devo tratar os números de benchmark?

Split the list. The coding jumps — Terminal-Bench 3.0 at 28.3, DeepSWE v1.1 at 66.9, SWE-Marathon at 42.5 — and the cyber results — CyberGym 84.5%, ExploitBench 54.4% — all come from Z.ai's own announcement and remain vendor-reported until an independent harness reproduces them. The Artificial Analysis Intelligence Index of 60 is the first independent measurement, and it is the number to weigh against everything Z.ai claims.

The leak was real, and the launch confirmed the name, the framing, and the timing — while correcting the one specific the rumor mill got loudest about. GLM-5.3 is the same base, post-trained hard, and now it carries an independent score to hold its vendor claims against: 60 on the Artificial Analysis Intelligence Index, tied with Kimi K3, seven ahead of GLM-5.2. The coding and cyber numbers are still Z.ai's own, the weights land on August 28, and the license line and a genuinely independent probe of the security claims are what's left to settle. Until then, the low-risk way to form your own view is a slice of real traffic and a failover to something proven.