release: 1.21.1 — re-sync Router Core to 3aca0e1 (Gemini 4 Argon capabilities)

CHANGELOG.md

@@ -2,6 +2,20 @@
 
 All notable changes to blockrun-llm will be documented in this file.
 
+## 1.21.1 — 2026-10-20
+
+### Changed
+- **Router Core re-synced to upstream `3aca0e1`** (was `2cd5a10`; the
+  TypeScript SDK moves to the same commit in 3.24.4). The only upstream change
+  is a `google/gemini-4-argon` capability entry — 1M context, 65,536 max
+  output, vision, `supports_tools` False until a paid tool probe verifies it,
+  so tool requests fail closed. The 88-decision parity fixture is re-copied
+  and byte-identical, so no routing decision changes.
+
+### Docs
+- README model list adds `google/gemini-4-argon` (Gemini 4 Argon, live on Base
+  and Solana since 2026-10-20). Catalog only: no routing tier picks it.
+
 ## 1.21.0 — 2026-10-20
 
 ### Changed

VERSION

@@ -1 +1 @@
-1.21.0
+1.21.1

blockrun_llm/__init__.py

@@ -196,7 +196,7 @@
     create_wallet as generate_wallet,  # User-friendly alias
 )
 
-__version__ = "1.21.0"
+__version__ = "1.21.1"
 __all__ = [
     "DEFAULT_API_KEY_URL",
     "ENV_API_KEY",

blockrun_llm/router_core/__init__.py

@@ -2,7 +2,7 @@
 Router Core — deterministic, constraint-first model routing.
 
 Python port of `@blockrun/router-core <https://github.com/BlockRunAI/router-core>`_
-(upstream commit ``2cd5a10``), the same routing engine the TypeScript SDK and
+(upstream commit ``3aca0e1``), the same routing engine the TypeScript SDK and
 the BlockRun gateway use. The package is deliberately product-neutral: task
 classification, hard capability filtering, portfolio scoring, ordered
 fallbacks, and routing configuration. It contains no wallet, gateway client,

blockrun_llm/router_core/config.py

@@ -2,7 +2,7 @@
 Default Routing Config
 
 Python port of ``@blockrun/router-core`` ``config.ts`` (upstream commit
-``2cd5a10``; the file is unchanged since ``e38958b``, 2026-09-29, so it is the
+``3aca0e1``; the file is unchanged since ``e38958b``, 2026-09-29, so it is the
 same table the TypeScript SDK bundles). V3.5: every tier rung is a public
 catalog id.
 

blockrun_llm/router_core/model_capabilities.py

@@ -200,6 +200,13 @@
             "supports_tools": True,
             "supports_vision": True,
         },
+        # supportsTools: not probed — fails closed
+        "google/gemini-4-argon": {
+            "context_window": 1_048_576,
+            "max_output_tokens": 65_536,
+            "supports_tools": False,
+            "supports_vision": True,
+        },
         "minimax/minimax-m2.7": {
             "context_window": 204_800,
             "max_output_tokens": 16_384,

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "blockrun-llm"
-version = "1.21.0"
+version = "1.21.1"
 description = "BlockRun SDK - Pay-per-request AI (LLM, Image, Video, Music, Voice) via x402 on Base and Solana"
 readme = "README.md"
 license = "MIT"

tests/unit/test_router_core.py

@@ -5,7 +5,7 @@
 case (``model-capabilities.test.ts``, ``portfolio.test.ts``, ``rules.test.ts``,
 ``selector.test.ts``, ``strategy.test.ts``, ``tokens.test.ts``,
 ``tool-intent.test.ts``, ``unavailable-models.test.ts`` at commit
-``2cd5a10``). They are the regression guard
+``3aca0e1``). They are the regression guard
 that the Python port keeps choosing the same models as the TypeScript SDK —
 when upstream is re-synced, re-port these alongside the source.
 """

tests/unit/test_router_core_snapshot.py

@@ -2,7 +2,7 @@
 Cross-language decision-snapshot parity.
 
 ``router_core_decisions.snapshot.json`` is a verbatim copy of upstream
-``decisions.snapshot.json`` at commit ``2cd5a10`` — 88 complete decisions the
+``decisions.snapshot.json`` at commit ``3aca0e1`` — 88 complete decisions the
 TypeScript engine produced for a frozen corpus (22 prompts x 4 profiles with
 rotating tool/vision/structured-output shapes, frozen pricing, frozen clock).
 This test recomputes every decision with the Python port and compares field