diff --git a/README.md b/README.md
index 0a5902a..d912615 100644
--- a/README.md
+++ b/README.md
@@ -6,10 +6,10 @@ An AI-powered Telegram bot playing "el gordo" — a blunt, politically incorrect
 
 ## Features
 
-- **AI chat**: configurable personality with web search, powered by Qwen via OpenRouter
+- **AI chat**: configurable personality with web search, powered by DeepSeek via OpenRouter
 - **Streaming responses**: AI replies stream token-by-token to Telegram (when no tools are active)
 - **Chat memory with RediSearch**: persistent conversation history, full-text search, automatic compaction
-- **Incremental summaries**: `/resumen` streams conversation summaries using Minimax, with automatic context compaction
+- **Incremental summaries**: `/resumen` streams conversation summaries using DeepSeek, with automatic context compaction
 - **Agentic tools**: AI can call tools (price lookup, calculator, web fetch, task scheduling) via function calling
 - **Market data**: `/prices`, `/usd`, `/petroleo`, `/devo`, `/powerlaw`, `/rainbow`, `/rulo`, `/eleccion`
 - **BCRA economic data**: `/bcra`, `/variables`
@@ -52,10 +52,10 @@ uv run --locked python run_polling.py
 
 | Use | Provider | Model |
 |---|---|---|
-| **Chat** | OpenRouter | `qwen/qwen3.6-plus` |
+| **Chat** | OpenRouter | `~deepseek/deepseek-v4-flash-latest` |
 | **Vision** | OpenRouter | `google/gemini-3.1-flash-lite-preview` |
 | **Transcription** | Groq → OpenRouter fallback | `whisper-large-v3` → `google/gemini-3.1-flash-lite-preview` |
-| **Summary** | OpenRouter | `minimax/minimax-m2.7` |
+| **Summary** | OpenRouter | `~deepseek/deepseek-v4-flash-latest` |
 
 **Streaming**: token streaming only when no tools/web-search are active. Tool-enabled requests return complete responses.
 
diff --git a/api/ai/pricing.py b/api/ai/pricing.py
index 41fd68f..736cca6 100644
--- a/api/ai/pricing.py
+++ b/api/ai/pricing.py
@@ -30,7 +30,7 @@
         "input_per_million": 250_000,
         "output_per_million": 1_500_000,
     },
-    "deepseek/deepseek-v4-flash": {
+    "~deepseek/deepseek-v4-flash-latest": {
         "input_per_million": 400_000,
         "output_per_million": 1_200_000,
     },
@@ -40,7 +40,7 @@
 def chat_output_token_limit(model: str) -> int:
     """Return a larger budget only for chat models that use hidden reasoning."""
 
-    if str(model or "").split(":", 1)[0] == "deepseek/deepseek-v4-flash":
+    if str(model or "").split(":", 1)[0] == "~deepseek/deepseek-v4-flash-latest":
         return REASONING_CHAT_OUTPUT_TOKEN_LIMIT
     return CHAT_OUTPUT_TOKEN_LIMIT
 
@@ -137,9 +137,9 @@ def estimate_chat_reserve_credits(
     messages: Sequence[Mapping[str, Any]],
     max_output_tokens: Optional[int] = None,
     extra_input_tokens: int = 0,
-    model: str = "deepseek/deepseek-v4-flash",
+    model: str = "~deepseek/deepseek-v4-flash-latest",
 ) -> int:
-    pricing = MODEL_PRICING_USD_MICROS.get(model, MODEL_PRICING_USD_MICROS["deepseek/deepseek-v4-flash"])
+    pricing = MODEL_PRICING_USD_MICROS.get(model, MODEL_PRICING_USD_MICROS["~deepseek/deepseek-v4-flash-latest"])
     output_token_limit = (
         chat_output_token_limit(model)
         if max_output_tokens is None
diff --git a/api/index.py b/api/index.py
index 9b8c0f0..d1fae30 100644
--- a/api/index.py
+++ b/api/index.py
@@ -325,8 +325,8 @@ def _parse_timeframe(
 
 def make_chat_tz(offset: int = -3) -> timezone:
     return timezone(timedelta(hours=offset))
-PRIMARY_CHAT_MODEL = "deepseek/deepseek-v4-flash"
-SUMMARY_MODEL = "deepseek/deepseek-v4-flash"
+PRIMARY_CHAT_MODEL = "~deepseek/deepseek-v4-flash-latest"
+SUMMARY_MODEL = "~deepseek/deepseek-v4-flash-latest"
 SUMMARY_MAX_TOKENS = 2048
 COMPACTION_THRESHOLD = 40
 COMPACTION_KEEP = 25
diff --git a/benchmark_bot.py b/benchmark_bot.py
index 6acb530..2b240fb 100644
--- a/benchmark_bot.py
+++ b/benchmark_bot.py
@@ -35,13 +35,13 @@ def __init__(self, api_key: str):
         self.models = [
             "qwen/qwen3.6-plus",
             "minimax/minimax-m2.7",
-            "deepseek/deepseek-v4-flash",
+            "~deepseek/deepseek-v4-flash-latest",
         ]
 
         self.model_pricing = {
             "qwen/qwen3.6-plus": {"input": 0.325, "output": 1.95, "context": "1M"},
             "minimax/minimax-m2.7": {"input": 0.30, "output": 1.20, "context": "196K"},
-            "deepseek/deepseek-v4-flash": {"input": 0.40, "output": 1.20, "context": "2M"},
+            "~deepseek/deepseek-v4-flash-latest": {"input": 0.40, "output": 1.20, "context": "1M"},
         }
 
     def load_bot_config(self) -> Dict[str, Any]:
diff --git a/tests/test_admin_reporting.py b/tests/test_admin_reporting.py
index cc94533..6af881f 100644
--- a/tests/test_admin_reporting.py
+++ b/tests/test_admin_reporting.py
@@ -68,7 +68,7 @@ def test_admin_report_formats_additional_context_in_english():
         index.app_runtime.admin.report(
             "OpenRouter unexpected finish_reason=None",
             extra_context={
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "enable_web_search": True,
             },
         )
@@ -76,7 +76,7 @@ def test_admin_report_formats_additional_context_in_english():
     assert send_msg.call_args.args[1] == (
         "admin report from VPS: OpenRouter unexpected finish_reason=None"
         "\n\nadditional context:"
-        "\nmodel: deepseek/deepseek-v4-flash"
+        "\nmodel: ~deepseek/deepseek-v4-flash-latest"
         "\nenable_web_search: True"
     )
 
diff --git a/tests/test_ai_billing_internal.py b/tests/test_ai_billing_internal.py
index c8ecfc9..c3670b1 100644
--- a/tests/test_ai_billing_internal.py
+++ b/tests/test_ai_billing_internal.py
@@ -48,11 +48,11 @@ def test_build_insufficient_credits_message_mentions_group_balances():
 
 def test_chat_output_token_limit_is_model_specific():
     assert (
-        chat_output_token_limit("deepseek/deepseek-v4-flash")
+        chat_output_token_limit("~deepseek/deepseek-v4-flash-latest")
         == REASONING_CHAT_OUTPUT_TOKEN_LIMIT
         == 8192
     )
-    assert chat_output_token_limit("deepseek/deepseek-v4-flash:exacto") == 8192
+    assert chat_output_token_limit("~deepseek/deepseek-v4-flash-latest:exacto") == 8192
     assert chat_output_token_limit("other/model") == CHAT_OUTPUT_TOKEN_LIMIT == 1024
 
 
@@ -76,7 +76,7 @@ def test_calculate_billing_for_segments_applies_cached_token_discount():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1_000,
                     "input_cached_tokens": 900,
@@ -92,7 +92,7 @@ def test_calculate_billing_for_segments_applies_cached_token_discount():
     assert breakdown["charged_credits_display"] == "0.2"
     assert breakdown["model_breakdown"] == [
         {
-            "model": "deepseek/deepseek-v4-flash",
+            "model": "~deepseek/deepseek-v4-flash-latest",
             "usd_micros": 1_000,
             "input_tokens": 1_000,
             "input_cached_tokens": 900,
@@ -107,12 +107,12 @@ def test_calculate_billing_for_segments_normalizes_billing_model_ids():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {"input_tokens": 100, "output_tokens": 50},
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {"input_tokens": 100, "output_tokens": 50},
             },
             {
@@ -135,8 +135,8 @@ def test_calculate_billing_for_segments_normalizes_billing_model_ids():
 
     assert breakdown["raw_usd_micros"] > 0
     assert [item["model"] for item in breakdown["model_breakdown"]] == [
-        "deepseek/deepseek-v4-flash",
-        "deepseek/deepseek-v4-flash",
+        "~deepseek/deepseek-v4-flash-latest",
+        "~deepseek/deepseek-v4-flash-latest",
         "google/gemini-3.1-flash-lite-preview",
         "google/gemini-3.1-flash-lite-preview",
         "groq/whisper-large-v3",
@@ -148,7 +148,7 @@ def test_calculate_billing_for_segments_bumps_pricing_version_for_deepseek_searc
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {"input_tokens": 100, "output_tokens": 50},
                 "metadata": {"web_search_requests": 1},
             }
@@ -163,7 +163,7 @@ def test_calculate_billing_for_segments_reads_cached_tokens_from_prompt_token_de
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "prompt_tokens": 2_000,
                     "completion_tokens": 100,
@@ -178,7 +178,7 @@ def test_calculate_billing_for_segments_reads_cached_tokens_from_prompt_token_de
     assert breakdown["charged_credits_display"] == "0.2"
     assert breakdown["model_breakdown"] == [
         {
-            "model": "deepseek/deepseek-v4-flash",
+            "model": "~deepseek/deepseek-v4-flash-latest",
             "usd_micros": 920,
             "input_tokens": 2_000,
             "input_cached_tokens": 1_500,
@@ -194,7 +194,7 @@ def test_calculate_billing_for_segments_skips_cached_source_segments():
             {
                 "kind": "chat",
                 "source": "cache",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 10_000,
                     "output_tokens": 500,
@@ -216,7 +216,7 @@ def test_calculate_billing_for_segments_bills_web_search_requests():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {"input_tokens": 100, "output_tokens": 50},
                 "metadata": {"web_search_requests": 2},
             }
@@ -235,7 +235,7 @@ def test_calculate_billing_for_segments_refunds_cache_only_usage_to_zero():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "source": "cache",
                 "usage": {
                     "input_tokens": 10_000,
@@ -288,7 +288,7 @@ def test_settle_reserved_ai_credits_refunds_successful_unused_reserve():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 100,
                     "output_tokens": 50,
@@ -318,7 +318,7 @@ def test_settle_reserved_ai_credits_charges_extra_when_actual_exceeds_reserve():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 4000,
                     "output_tokens": 3000,
@@ -404,7 +404,7 @@ def test_settle_reserved_ai_credits_records_debt_when_extra_charge_fails():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 4000,
                     "output_tokens": 3000,
@@ -447,7 +447,7 @@ def test_settle_reserved_ai_credits_batch_converts_to_credits_once_and_refunds_o
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "output_tokens": 1,
@@ -492,7 +492,7 @@ def test_settle_reserved_ai_credits_batch_mixed_sources_refunds_later_reserves()
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 100,
                     "output_tokens": 50,
@@ -606,7 +606,7 @@ def test_settle_reserved_ai_credits_batch_charges_extra_once_when_total_exceeds_
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 4000,
                     "output_tokens": 2000,
@@ -635,7 +635,7 @@ def test_settle_reserved_ai_credits_keeps_reserve_when_groq_reports_zero_usage()
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 0,
                     "output_tokens": 0,
@@ -669,7 +669,7 @@ def test_settle_reserved_ai_credits_refunds_cache_only_usage():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "prompt_tokens": 1_000,
                     "prompt_tokens_details": {"cached_tokens": 900},
@@ -712,7 +712,7 @@ def test_settle_reserved_ai_credits_batch_keeps_full_reserve_when_total_usage_is
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 0,
                     "output_tokens": 0,
@@ -763,7 +763,7 @@ def test_settle_reserved_ai_credits_batch_ignores_none_segments():
             None,
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 100,
                     "output_tokens": 50,
@@ -839,7 +839,7 @@ def test_settle_reserved_ai_credits_refunds_partial_chat_usage():
     segments = [
         {
             "kind": "chat",
-            "model": "deepseek/deepseek-v4-flash",
+            "model": "~deepseek/deepseek-v4-flash-latest",
             "usage": {
                 "input_tokens": 1_000,
                 "output_tokens": 100,
@@ -929,7 +929,7 @@ def test_calculate_billing_uses_gateway_cost_when_higher_than_local():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "prompt_tokens": 100,
                     "completion_tokens": 50,
@@ -951,7 +951,7 @@ def test_calculate_billing_keeps_local_cost_when_higher_than_gateway():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "prompt_tokens": 4_000,
                     "completion_tokens": 2_000,
@@ -970,7 +970,7 @@ def test_calculate_billing_without_gateway_cost_uses_local():
         [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "prompt_tokens": 100,
                     "completion_tokens": 50,
diff --git a/tests/test_ai_requests.py b/tests/test_ai_requests.py
index e941aa1..ee9eb0c 100644
--- a/tests/test_ai_requests.py
+++ b/tests/test_ai_requests.py
@@ -118,7 +118,7 @@ def test_log_groq_request_result_logs_local_billing_details():
     result = AIUsageResult(
         kind="chat",
         text="respuesta",
-        model="deepseek/deepseek-v4-flash",
+        model="~deepseek/deepseek-v4-flash-latest",
         usage={"input_tokens": 100, "output_tokens": 50},
         metadata={"groq_account": "primary"},
     )
@@ -183,7 +183,7 @@ def fake_attempt(account):
         return AIUsageResult(
             kind="chat",
             text="respuesta chat",
-            model="deepseek/deepseek-v4-flash",
+            model="~deepseek/deepseek-v4-flash-latest",
             metadata={"groq_account": account},
         )
 
@@ -548,7 +548,7 @@ def test_complete_with_providers_openrouter_success():
     openrouter_result = AIUsageResult(
         kind="chat",
         text="OpenRouter response",
-        model="deepseek/deepseek-v4-flash",
+        model="~deepseek/deepseek-v4-flash-latest",
         usage={"input_tokens": 100, "output_tokens": 50},
         metadata={"provider": "openrouter"},
     )
@@ -597,7 +597,7 @@ def test_complete_with_providers_records_openrouter_billing_on_success(monkeypat
     openrouter_result = AIUsageResult(
         kind="chat",
         text="OpenRouter response",
-        model="deepseek/deepseek-v4-flash",
+        model="~deepseek/deepseek-v4-flash-latest",
         usage={"input_tokens": 100, "output_tokens": 50},
         metadata={"provider": "openrouter"},
     )
diff --git a/tests/test_message_ai_media.py b/tests/test_message_ai_media.py
index ce54477..35a0e90 100644
--- a/tests/test_message_ai_media.py
+++ b/tests/test_message_ai_media.py
@@ -105,7 +105,7 @@ def test_handle_msg_image_conversation_with_two_provider_requests_reserves_base_
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
@@ -114,7 +114,7 @@ def test_handle_msg_image_conversation_with_two_provider_requests_reserves_base_
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
@@ -175,7 +175,7 @@ def test_handle_msg_image_conversation_settles_in_single_batch(monkeypatch):
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {"input_tokens": 1, "output_tokens": 1},
             },
         ],
@@ -416,7 +416,7 @@ def test_handle_msg_ai_flow_settles_with_single_base_reserve_when_usage_is_tiny(
         billing_segments=[
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
@@ -425,7 +425,7 @@ def test_handle_msg_ai_flow_settles_with_single_base_reserve_when_usage_is_tiny(
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
@@ -524,7 +524,7 @@ def test_handle_msg_ai_flow_keeps_single_reserve_for_three_tiny_segments(monkeyp
         billing_segments=[
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
@@ -533,7 +533,7 @@ def test_handle_msg_ai_flow_keeps_single_reserve_for_three_tiny_segments(monkeyp
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
@@ -542,7 +542,7 @@ def test_handle_msg_ai_flow_keeps_single_reserve_for_three_tiny_segments(monkeyp
             },
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
diff --git a/tests/test_message_billing.py b/tests/test_message_billing.py
index 6ff8cc6..1163abc 100644
--- a/tests/test_message_billing.py
+++ b/tests/test_message_billing.py
@@ -449,14 +449,14 @@ def test_handle_msg_creditlog_admin_shows_recent_settlements(monkeypatch):
                     "raw_usd_micros": 390,
                     "model_breakdown": [
                         {
-                            "model": "deepseek/deepseek-v4-flash",
+                            "model": "~deepseek/deepseek-v4-flash-latest",
                             "usd_micros": 325,
                             "input_tokens": 1000,
                             "input_cached_tokens": 800,
                             "input_non_cached_tokens": 200,
                         },
                         {
-                            "model": "deepseek/deepseek-v4-flash",
+                            "model": "~deepseek/deepseek-v4-flash-latest",
                             "usd_micros": 65,
                             "input_tokens": 200,
                             "input_cached_tokens": 100,
@@ -499,7 +499,7 @@ def test_handle_msg_creditlog_admin_shows_recent_settlements(monkeypatch):
     assert "requests: chat=3" in sent_text
     assert "cache_hits: chat=1" in sent_text
     assert "cacheados=900 ahorro_cache=0" in sent_text
-    assert "deepseek/deepseek-v4-flash=390" in sent_text
+    assert "~deepseek/deepseek-v4-flash-latest=390" in sent_text
     assert "web_search=8000 (2x)" in sent_text
     assert "python=500 (1x)" in sent_text
 
diff --git a/tests/test_message_transcription.py b/tests/test_message_transcription.py
index ea44d73..1531875 100644
--- a/tests/test_message_transcription.py
+++ b/tests/test_message_transcription.py
@@ -626,7 +626,7 @@ def test_handle_msg_auto_audio_plus_ai_response_charges_three_requests(monkeypat
         billing_segments=[
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {
                     "input_tokens": 1,
                     "input_non_cached_tokens": 1,
diff --git a/tests/test_provider_runtime.py b/tests/test_provider_runtime.py
index 0ee0dfb..8098802 100644
--- a/tests/test_provider_runtime.py
+++ b/tests/test_provider_runtime.py
@@ -764,7 +764,7 @@ def _build_retry_runtime(responses, *, extract_usage=lambda _response: {}):
                 metadata=kwargs.get("metadata") or {},
             ),
             extract_usage_map=extract_usage,
-            primary_model="deepseek/deepseek-v4-flash",
+            primary_model="~deepseek/deepseek-v4-flash-latest",
             max_tool_rounds=5,
         ),
         ToolRuntime(),
@@ -825,7 +825,7 @@ def test_provider_runtime_retries_invalid_finish_reason_then_returns_result(
     assert result.text == "done"
     assert len(client.calls) == 2
     assert all(
-        call["max_tokens"] == chat_output_token_limit("deepseek/deepseek-v4-flash")
+        call["max_tokens"] == chat_output_token_limit("~deepseek/deepseek-v4-flash-latest")
         for call in client.calls
     )
     assert request_count.call_count == 2
@@ -908,7 +908,7 @@ def test_provider_runtime_reports_null_finish_reason_after_retries_exhausted():
     assert admin_report.call_args.args[0] == "OpenRouter unexpected finish_reason=None"
     report_context = admin_report.call_args.kwargs["extra_context"]
     assert report_context == {
-        "model": "deepseek/deepseek-v4-flash",
+        "model": "~deepseek/deepseek-v4-flash-latest",
         "enable_web_search": True,
         "tool_round": 1,
         "response_id": "gen-4",
diff --git a/tests/test_world_cup_goals.py b/tests/test_world_cup_goals.py
index 95b7da5..86d5f5c 100644
--- a/tests/test_world_cup_goals.py
+++ b/tests/test_world_cup_goals.py
@@ -367,7 +367,7 @@ def ask_ai(_messages, *, response_meta, **_kwargs):
         response_meta["billing_segments"] = [
             {
                 "kind": "chat",
-                "model": "deepseek/deepseek-v4-flash",
+                "model": "~deepseek/deepseek-v4-flash-latest",
                 "usage": {"prompt_tokens": 100, "completion_tokens": 50},
             }
         ]