SambaNova Brings GLM-5.3 to SambaCloud at Over 200 Tokens Per Second

SambaNova Systems said on November 11, 2026, that Z.ai’s open-weight GLM-5.3 model is available on its SambaCloud inference service at more than 200 tokens per second. Developers can access the model through the SambaCloud playground, an API, and integrations for coding agents.

Z.ai introduced GLM-5.3 on August 14, 2026, using the same base model as GLM-5.2 and attributing the newer system’s gains to additional post-training across more environments, more diverse tasks, and more compute. The SambaNova announcement adds a hosted deployment for that existing model. Its intended work includes complex software engineering and long-horizon agent tasks involving sequences of dependent steps.

What the Agent Trace Shows

SambaNova supplied a trace from an example agent run on SambaCloud to illustrate those demands. The run made 41 model calls over 17 minutes and 30 seconds, processing 5.4 million input tokens while generating just under 42,000 output tokens. Because each turn resent the accumulated context, input grew from 38,000 tokens on the first call to more than 114,000 by the eighth. Outputs remained small: a few hundred tokens containing reasoning and a tool call, and more than 91% of the input was served from cache.

That pattern explains the emphasis on serving speed. Agent loops repeatedly send long contexts back to the model while requesting only a short next action, so per-call latency accumulates across the sequence. SambaNova presents its reported throughput as a way to keep those steps short, while the cache figure demonstrates how heavily the workload reuses earlier context.

Post-Training and Reported Results

Z.ai says its post-training environments were designed to resemble professional work, including tasks that could occupy an experienced engineer for several days. In one machine-learning infrastructure scenario, the model receives access to compute clusters, storage systems, internal documentation, codebases, and experiment results. It must diagnose bottlenecks across a training stack, implement optimizations, run experiments, and deliver a measurable end-to-end speedup while preserving correctness.

Z.ai reports that GLM-5.3 improved from 4.6 to 28.3 on Terminal-Bench 3.0, from 46.2 to 66.9 on DeepSWE v1.1, and from 23.8 to 28.5 on Agents’ Last Exam. The comparisons place the gains in terminal work, software engineering, and general agent tasks, the same categories SambaNova is targeting with this deployment.

On its private Code Bench, Z.ai reports GLM-5.3 reaching 31.4% at High effort with about 50,000 output tokens per task, compared with Claude Opus 4.8 at 29.5% and 120,000 tokens. GLM-5.3 remained behind Claude Fable 5, which reached 39.5% at Max effort. That is a specific efficiency result: GLM-5.3 generated fewer tokens and scored higher than Opus 4.8 in this test, but Z.ai’s own comparison still places Fable 5 ahead.

API and Coding-Agent Access

On SambaCloud, the model is called with the model string GLM-5.3. SambaNova’s Python example sends chat-completion requests to https://api.sambanova.ai/v1 and sets reasoning_effort to max, the level Z.ai recommends for coding. Developers can use an API key or test the model in the SambaCloud playground.

ZCode can use SambaCloud through OpenAI- or Anthropic-compatible provider settings, with a base URL, API key, and GLM-5.3 selected in the model menu. SambaNova says the same base-URL change works with Claude Code, OpenCode, and Codex CLI. For Codex, developers are instructed to use the Responses API endpoint and declare the model’s supported reasoning levels in their model catalog.

GLM-5.3 supports low, high, and max reasoning effort, and thinking cannot be disabled. Z.ai says applications still using thinking.type: "disabled" must enable thinking and set reasoning effort to low before updating the model ID; otherwise, requests will fail.

For teams evaluating hosted agent inference, the distinction matters: benchmark scores describe what a model can complete, while the trace exposes the repeated serving work required to complete it. SambaNova’s deployment is aimed at that second part of the problem, making per-call speed and cache reuse as relevant to the user experience as GLM-5.3’s reported task scores.