harn-vm 0.10.136

Async bytecode virtual machine for the Harn programming language
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61

# ---------- DeepSeek V4 direct (OpenAI-compatible /v1) ------------------------
# V4-Flash + V4-Pro support native tool calls, thinking mode, prompt
# caching, and JSON output. DeepSeek's V4 notice retires
# `deepseek-chat`/`deepseek-reasoner` on 2026-07-24 15:59 UTC, so the
# direct catalog intentionally exposes only V4 model IDs.

[[provider.deepseek]]
model_match = "deepseek-v4-pro*"
native_tools = true
preferred_tool_format = "native"
structured_output = "native"
thinking_modes = ["enabled", "effort"]
reasoning_effort_supported = true
reasoning_effort_levels = ["low", "high", "max"]
reasoning_disable_supported = true
reasoning_wire_format = "thinking_type"
reasoning_history_wire_field = "reasoning_content"
reasoning_round_trip = "echo_same_key_when_tools_present"
reasoning_excluded_portable_options = [
  "temperature",
  "top_p",
  "presence_penalty",
  "frequency_penalty",
]
text_tool_wire_format_supported = true
prefers_xml_scaffolding = false
prefers_markdown_scaffolding = true
structured_output_mode = "native_json"
supports_assistant_prefill = false
prefers_role_developer = false
prefers_xml_tools = false
thinking_block_style = "inline"
prompt_caching = true

[[provider.deepseek]]
model_match = "deepseek-v4-flash*"
native_tools = true
preferred_tool_format = "native"
structured_output = "native"
thinking_modes = ["enabled", "effort"]
reasoning_effort_supported = true
reasoning_effort_levels = ["low", "high", "max"]
reasoning_disable_supported = true
reasoning_wire_format = "thinking_type"
reasoning_history_wire_field = "reasoning_content"
reasoning_round_trip = "echo_same_key_when_tools_present"
reasoning_excluded_portable_options = [
  "temperature",
  "top_p",
  "presence_penalty",
  "frequency_penalty",
]
text_tool_wire_format_supported = true
prefers_xml_scaffolding = false
prefers_markdown_scaffolding = true
structured_output_mode = "native_json"
supports_assistant_prefill = false
prefers_role_developer = false
prefers_xml_tools = false
thinking_block_style = "inline"
prompt_caching = true