Skip to content

feat(code): replace Gemini 3.5 Flash with 3.6 Flash in model switcher #4890

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Mason Daugherty (mdrxy) merged 2 commits into from
Jul 21, 2026
Merged

feat(code): replace Gemini 3.5 Flash with 3.6 Flash in model switcher #4890

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ee85810
feat(code): replace Gemini 3.5 Flash with 3.6 Flash in model switcher
mdrxy Jul 21, 2026
237955f
Merge branch 'main' into mdrxy/code/gemini-3-6-flash
mdrxy Jul 21, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions libs/code/deepagents_code/reasoning_effort.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ class ReasoningProviderConfig(NamedTuple):
"""Gemini `thinking_level` labels.

Applied to every `gemini-3*` model (the gate in `_classify_reasoning_provider`),
including Gemini 3 Pro/Flash, 3.1 Pro, and 3.5 Flash — all accept
including Gemini 3 Pro/Flash, 3.1 Pro, and 3.6 Flash — all accept
low/medium/high. `minimal` is Flash-Lite / original-Pro territory, neither of
which is offered here. See https://ai.google.dev/gemini-api/docs/thinking.
"""
Expand Down Expand Up @@ -278,7 +278,7 @@ def _google_supported_efforts(_model: str) -> tuple[EffortLabel, ...]:

def _google_default_effort(model: str) -> EffortLabel | None:
"""Return the Gemini default thinking level when known."""
if model.startswith("gemini-3.5-flash"):
if model.startswith("gemini-3.6-flash"):
return "medium"
if model.startswith(("gemini-3.1-pro", "gemini-3-flash", "gemini-3-pro")):
return "high"
Expand Down
4 changes: 2 additions & 2 deletions libs/code/deepagents_code/tui/widgets/model_selector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,7 +77,7 @@
"fireworks:accounts/fireworks/models/kimi-k2p7-code": "Kimi K2.7 Code",
"fireworks:accounts/fireworks/models/minimax-m3": "MiniMax-M3",
"fireworks:accounts/fireworks/models/qwen3p7-plus": "Qwen 3.7 Plus",
"google_genai:gemini-3.5-flash": "Gemini 3.5 Flash",
"google_genai:gemini-3.6-flash": "Gemini 3.6 Flash",
"google_genai:gemini-3.1-pro-preview": "Gemini 3.1 Pro Preview",
"meta:muse-spark-1.1": "Muse Spark 1.1",
"ollama:deepseek-v4-flash:cloud": "DeepSeek V4 Flash",
Expand Down Expand Up @@ -109,7 +109,7 @@
"openrouter:deepseek/deepseek-v4-flash": "DeepSeek V4 Flash",
"openrouter:deepseek/deepseek-v4-flash:free": "DeepSeek V4 Flash (free)",
"openrouter:deepseek/deepseek-v4-pro": "DeepSeek V4 Pro",
"openrouter:google/gemini-3.5-flash": "Gemini 3.5 Flash",
"openrouter:google/gemini-3.6-flash": "Gemini 3.6 Flash",
"openrouter:google/gemini-3.1-pro-preview": "Gemini 3.1 Pro Preview",
"openrouter:moonshotai/kimi-k2.7-code": "Kimi K2.7 Code",
"openrouter:moonshotai/kimi-k3": "Kimi K3",
Expand Down
2 changes: 1 addition & 1 deletion libs/code/tests/unit_tests/test_configurable_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -622,7 +622,7 @@ def test_existing_headers_preserved_and_session_affinity_not_overwritten(
}

def test_non_fireworks_non_openai_model_unchanged_with_thread_id(self) -> None:
model = _make_model("gemini-3.5-flash")
model = _make_model("gemini-3.6-flash")
model._get_ls_params.return_value = {"ls_provider": "google_genai"}
request = _make_request(
model,
Expand Down
12 changes: 6 additions & 6 deletions libs/code/tests/unit_tests/test_reasoning_effort.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ def _restore_settings(
# single-digit version it prefixes (`opus-4-1` must not match
# `opus-4-16`); it falls through to the full range instead.
("anthropic:claude-opus-4-16", ("low", "medium", "high", "xhigh", "max")),
("google_genai:gemini-3.5-flash", ("low", "medium", "high")),
("google_genai:gemini-3.6-flash", ("low", "medium", "high")),
("google_genai:gemini-3.1-pro-preview", ("low", "medium", "high")),
("xai:grok-4.5", ("low", "medium", "high")),
("xai:grok-4.5-latest", ("low", "medium", "high")),
Expand Down Expand Up @@ -140,7 +140,7 @@ def test_supported_efforts_for_model(model_spec: str, efforts: tuple[str, ...])
("anthropic:claude-sonnet-4-6", "high"),
("anthropic:claude-sonnet-4-5", None),
("anthropic:claude-opus-4-1", None),
("google_genai:gemini-3.5-flash", "medium"),
("google_genai:gemini-3.6-flash", "medium"),
("google_genai:gemini-3.1-pro-preview", "high"),
("google_genai:gemini-3-pro", "high"),
("google_genai:gemini-3-flash", "high"),
Expand Down Expand Up @@ -168,7 +168,7 @@ def test_model_params_for_effort_maps_provider_kwargs() -> None:
"thinking": {"type": "adaptive", "display": "summarized"},
"output_config": {"effort": "xhigh"},
}
assert model_params_for_effort("google_genai:gemini-3.5-flash", "low") == {
assert model_params_for_effort("google_genai:gemini-3.6-flash", "low") == {
"thinking_level": "low"
}
assert model_params_for_effort(
Expand Down Expand Up @@ -253,7 +253,7 @@ def test_merge_and_clear_xai_effort_preserves_extra_body_params() -> None:
("openai:gpt-5.5", {"reasoning": {"effort": 5}}),
("anthropic:claude-opus-4-8", {"output_config": {"effort": 5}}),
("anthropic:claude-opus-4-8", {"output_config": "high"}),
("google_genai:gemini-3.5-flash", {"thinking_level": 5}),
("google_genai:gemini-3.6-flash", {"thinking_level": 5}),
(
"fireworks:accounts/fireworks/models/deepseek-v4-pro",
{"model_kwargs": {"reasoning_effort": 5}},
Expand Down Expand Up @@ -710,7 +710,7 @@ def test_without_effort_clears_legacy_anthropic_top_level_effort() -> None:


def test_without_effort_clears_google_thinking_level() -> None:
effort_params = model_params_for_effort("google_genai:gemini-3.5-flash", "low")
effort_params = model_params_for_effort("google_genai:gemini-3.6-flash", "low")
assert effort_params is not None
assert without_effort_model_params(effort_params) is None

Expand All @@ -736,7 +736,7 @@ def test_without_effort_preserves_non_dict_model_kwargs() -> None:
("openai:gpt-5.5", "none"),
("openai:gpt-5.5", "high"),
("anthropic:claude-opus-4-8", "xhigh"),
("google_genai:gemini-3.5-flash", "low"),
("google_genai:gemini-3.6-flash", "low"),
("fireworks:accounts/fireworks/models/deepseek-v4-pro", "max"),
],
)
Expand Down