Skip to content
Python

Nebius: two verified rows with windows, tail of the gallery

Rohit C Prasad7a3c14ec21fe
diff --git a/coworker/providers/matrix.py b/coworker/providers/matrix.py
index bc1dca42..1d1ed5ee 100644
--- a/coworker/providers/matrix.py
+++ b/coworker/providers/matrix.py
@@ -381,12 +381,13 @@ MATRIX: dict[str, ModelEntry] = {
"vertex:openweight/qwen/qwen3-coder-480b-a35b-instruct-maas": ModelEntry(
"Qwen3 Coder · Vertex AI", _AGENTIC, 256_000
),
- # Nebius Token Factory (ids verified against /v1/models + a live tool-call round-trip
- # on 2026-07-26; it keeps each lab's own namespace, like Together).
- "nebius:zai-org/GLM-5.2": ModelEntry("GLM-5.2 · via Nebius"),
- "nebius:deepseek-ai/DeepSeek-V4-Pro": ModelEntry("DeepSeek V4 Pro · via Nebius"),
- "nebius:moonshotai/Kimi-K2.6": ModelEntry("Kimi K2.6 · via Nebius"),
- "nebius:Qwen/Qwen3.5-397B-A17B": ModelEntry("Qwen3.5 397B · via Nebius"),
+ # Nebius Token Factory keeps each lab's own namespace, like Together. Ids and windows
+ # from its public catalogue (tokenfactory.nebius.com/api/public/models_info, read
+ # 2026-10-25: `max_model_len` 1,048,576 for both; function_calling listed).
+ "nebius:zai-org/GLM-5.2": ModelEntry("GLM-5.2 · via Nebius", _AGENTIC, 1_048_576),
+ "nebius:deepseek-ai/DeepSeek-V4-Pro": ModelEntry(
+ "DeepSeek V4 Pro · via Nebius", _AGENTIC, 1_048_576
+ ),
}
diff --git a/coworker/providers/registry.py b/coworker/providers/registry.py
index acdf677b..e74dda94 100644
--- a/coworker/providers/registry.py
+++ b/coworker/providers/registry.py
@@ -735,7 +735,6 @@ DESCRIPTORS: list[ProviderDescriptor] = [
base_url="https://api.tokenfactory.nebius.com/v1",
recommended_model="zai-org/GLM-5.2",
env_key="NEBIUS_API_KEY",
- endpoint_help="Prefilled with Nebius Token Factory's endpoint (the former AI Studio host, https://api.studio.nebius.com/v1, still resolves).",
),
ProviderDescriptor(
name="ollama",
diff --git a/surfaces/gui/src/providers/logos.ts b/surfaces/gui/src/providers/logos.ts
index db042f10..fa670d7a 100644
--- a/surfaces/gui/src/providers/logos.ts
+++ b/surfaces/gui/src/providers/logos.ts
@@ -75,7 +75,6 @@ export const PROVIDER_ORDER = [
"openrouter-account",
"fireworks",
"together",
- "nebius",
"zai",
"kimi",
"deepseek",
@@ -83,6 +82,7 @@ export const PROVIDER_ORDER = [
"qwen",
"minimax",
"xai",
+ "nebius",
];
export function providerRank(name: string): number {
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 8311b6da..5c2a10a2 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -498,7 +498,7 @@ def test_matrix_answers_capabilities_for_reseller_ids():
"fireworks:accounts/fireworks/models/kimi-k2p6",
"openrouter:z-ai/glm-5.2",
"openrouter:meta-llama/llama-4-maverick",
- "nebius:Qwen/Qwen3.5-397B-A17B",
+ "nebius:zai-org/GLM-5.2",
):
caps = capabilities_for(mid)
assert caps.tools and caps.parallel_tool_calls and caps.streaming