hugo/DS4Server
1
0
Fork 0
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

Remove deprecated Flash preview and legacy MTP

Browse Source
This commit is contained in:
Georg Bauer
2026-08-31 08:04:20 +02:00
parent 7f88e23884
commit 0fa15bb68b
24 changed files with 246 additions and 1200 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
scripts/endpoint_continuation.py
View File

@@ -45,7 +45,7 @@ def post(path, payload):
first = post(
"/v1/chat/completions",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [{"role": "user", "content": prompt}],
"tools": [chat_tool],
"reasoning_effort": "none",
@@ -58,7 +58,7 @@ print("chat-first", json.dumps(first["usage"], separators=(",", ":")))
second = post(
"/v1/chat/completions",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [
{"role": "user", "content": prompt},
{"role": "assistant", "content": "", "tool_calls": [call]},
@@ -79,7 +79,7 @@ print("chat", json.dumps(second, separators=(",", ":")))
first = post(
"/v1/messages",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [{"role": "user", "content": prompt}],
"tools": [anthropic_tool],
"thinking": {"type": "disabled"},
@@ -92,7 +92,7 @@ print("anthropic-first", json.dumps(first["usage"], separators=(",", ":")))
second = post(
"/v1/messages",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [
{"role": "user", "content": prompt},
{"role": "assistant", "content": [call]},
@@ -118,7 +118,7 @@ print("anthropic", json.dumps(second, separators=(",", ":")))
first = post(
"/v1/responses",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"input": prompt,
"tools": [responses_tool],
"reasoning": {"effort": "none"},
@@ -131,7 +131,7 @@ print("responses-first", json.dumps(first["usage"], separators=(",", ":")))
second = post(
"/v1/responses",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"input": [
{
"type": "message",

6
scripts/endpoint_parity.py
View File

@@ -16,7 +16,7 @@ cases = [
"chat",
"/v1/chat/completions",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [
{
"role": "user",
@@ -42,7 +42,7 @@ cases = [
"anthropic",
"/v1/messages",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [
{
"role": "user",
@@ -65,7 +65,7 @@ cases = [
"responses",
"/v1/responses",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"input": "Call echo with text hi. Do not answer normally.",
"tools": [
{

6
scripts/endpoint_reasoning.py
View File

@@ -11,7 +11,7 @@ cases = [
"chat",
"/v1/chat/completions",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [{"role": "user", "content": prompt}],
"reasoning_effort": "low",
"temperature": 0,
@@ -22,7 +22,7 @@ cases = [
"anthropic",
"/v1/messages",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [{"role": "user", "content": prompt}],
"thinking": {"type": "enabled", "budget_tokens": 64},
"output_config": {"effort": "low"},
@@ -34,7 +34,7 @@ cases = [
"responses",
"/v1/responses",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"input": prompt,
"reasoning": {"effort": "low", "summary": "auto"},
"temperature": 0,

3
scripts/execution_parity.py
View File

@@ -12,9 +12,8 @@ HARDWARE_TESTS = (
"flash_resident_and_ssd_streaming_choose_the_same_tokens",
"flash_ssd_streaming_maps_batched_prefill_layers",
"resident_multi_session_switching_preserves_each_kv_frontier",
"legacy_mtp_runs_a_target_owned_greedy_cycle",
"dspark_runs_a_target_owned_greedy_cycle",
"ssd_streaming_supports_legacy_mtp_and_dspark",
"ssd_streaming_supports_dspark",
"directional_steering_matches_the_ds4_token_oracle",
"resident_and_streamed_glm_match_ds4_decode_oracles",
"streamed_glm_uses_ds4_indexed_prefill_for_long_prompts",

6
scripts/speculative_parity.py
View File

@@ -28,7 +28,7 @@ def chat(base_url, prompt):
base_url,
"/v1/chat/completions",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [{"role": "user", "content": prompt}],
"reasoning_effort": "none",
"temperature": 0,
@@ -50,7 +50,7 @@ def anthropic(base_url, prompt):
base_url,
"/v1/messages",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"messages": [{"role": "user", "content": prompt}],
"thinking": {"type": "disabled"},
"temperature": 0,
@@ -69,7 +69,7 @@ def responses(base_url, prompt):
base_url,
"/v1/responses",
{
"model": "deepseek-v4-flash",
"model": "deepseek-v4-flash-0731",
"input": prompt,
"reasoning": {"effort": "none"},
"temperature": 0,