The word translated as "forever" can be interpreted "long time, period, duration, farthest reaches of time, age".
There does not exist a word in Hebrew that unambiguously means "forever".
Moreover the meaning of this word changes over time in Hebrew as we start seeing plural forms, suggesting that olam shifted to mean "age".
There are many interpretations of this word, and especially when reading the later texts of Chronicles, Ezra, Nehemiah and the later prophets, I would not recommend translating it as "forever". See Appendix.
Some are claiming that different language was used to describe different covenants (e.g. Sinai, Abraham, etc.). This is incorrect. The same 'olam berith' is used to describe all five covenants (with Noah, Abraham, Moses, Aaron, and David).
- With Noah: Gen 9.16 (LEB)
The bow shall be in the clouds, and I will see it, so as to remember
the everlasting covenant ("olam berith") between God and between
every living creature, with all flesh that is upon the earth.”
- With Abraham and his descendents
Gen 17.7 (LEB)
And I will establish my covenant between me and you, and between your
offspring after you, throughout their generations as an everlasting
covenant ("olam berith") to be as God for you and to your offspring
- With the Israelites at Sinai
Ex 31.16-17 (LEB)
The Israelites will pay attention to the Sabbath in order to fulfill
the Sabbath throughout their generations as a lasting covenant
("olam berith"). It is a sign between me and the Israelites
forever ("olam"), because in six days Yahweh made the heavens and the earth, and on the seventh he ceased and recovered.”
Numbers 18.19 (LEB)
All the contributions of holiness that the Israelites offer to Yahweh
I have given to you and your sons and your daughters with you as an
eternal decree; it is an eternal covenant ("olam berith") of salt
before Yahweh to you and your offspring with you.”
Leviticus 24.8-9 (LEB)
On every Sabbath he shall arrange it in rows before Yahweh
continually; they are from the Israelites as an everlasting
covenant ("olam berith"). And it shall be for Aaron and for his
sons, and they shall eat it in a holy place, because it is a most holy
thing for him from Yahweh’s offerings made by fire—a lasting rule.
- with Aaron and his descendents
Yahweh spoke to Moses, saying, “Phinehas son of Eleazar, son of Aaron
the priest, turned away my anger from among the Israelites when he was
jealous with my jealousy in their midst, and I did not destroy the
Israelites with my jealousy. Therefore say, ‘Behold, I am giving to
him my covenant of peace, and it will be for him and his offspring
after him a covenant of an eternal priesthood ("berith olam
kehunnath") because he was jealous for his God and made atonement for
the Israelites.’ ”
- With David and his descendants
2 Sam 23.5 (LEB)
Yet not so is my house with God, for he made an everlasting
covenant ("olam berith") for me, arranging everything. He has
secured all my deliverance, and all my desire he will cause to happen.
Given that claims that different language was used with Moses or Abraham is so easy to disprove (even a simple concordance will do) I really wonder why people keep making this argument, yet it so often comes up that, for example, it's OK that animals are no longer sacrificed but we still need to keep the sabbath. Or that we need to keep the feasts but it's fine if the levitical purity laws are no longer kept. Or that the covenant of Abraham was forever but the covenant of Moses wasn't, etc.
The same language was used for all of these, so if you want to argue that it's OK to do something but not another, you can still do so with other arguments, but not on the basis of 'olam berith' being used in some places but not others.
From the Theological Dictionary of the Old Testament:
Following E. Jenni, most scholars translate ʿôlām as “long time” or
“farthest, remotest time.” The various nuances of this translation
must then also be distinguished contextually.[...]
That ʿôlām can have a plural at all might indicate that it can also
refer to a “period of time,” with the plural then referring to
“periods of time, ages,” or something similar. This situation,
however, applies only to its later use in early Judaism and in the
following period. A genuinely numeric plural occurs in the Hebrew OT
at most only in Eccl. 1:10; otherwise the reference is usually to the
iterative, extensive, amplifying plural. What the few plural
occurrences of ʿôlām already show becomes quite evident in the
numerous occurrences of the singular. In the OT (as also in Ugaritic
texts), ʿôlām is used not as an independent subject or object but
rather largely within construct combinations or as an adverbial
accusative. Hence ʿôlām occurs in connection with terms for love (Jer.
31:3), signs (Isa. 55:13), joy (Isa. 35:10, etc.), shame and disgrace
(Ps. 78:66; Jer. 23:40), a heap of ruins (Dt. 13:17; Josh. 8:28),
appointments (Ex. 29:28; 30:21; Lev. 6:11, 15[18, 22] etc.),
possessions (Gen. 17:8; 48:4, etc.), berîṯ (16 times), etc. Similarly
frequent combinations with other future-oriented lexemes underscore
that ʿôlām functions to express the highest possible intensification
(“perpetual holding,” “unending joy,” etc.); in such combinations with
ʿôlām, these lexemes are themselves intensified (cf., e.g., the
combination with → עד ʿaḏ, with → חיים ḥayyîm, or with → דור dôr,
including examples in the pl.).13 Although ʿôlām is not yet attested
in extrabiblical Hebrew witnesses (ostraca, inscriptions), its
corresponding equivalents occur relatively frequently in texts within
the OT environs.14 Reference can be made first to Ugaritic witnesses,
particularly since combinations with the preps. le and ʿaḏ are also
already attested here. Within Old Aramaic, texts from Sefire and
Ahiqar are joined by one witness from Deir ʿAlla.16 Occurrences in
Phoenician are frequent, while the orthography אולם is found in Punic.
The findings in Old South Arabic are disputed.18 The Moabite of the
Mesha inscription attests both meanings for ʿlm: “for always,
perpetual” and “since time immemorial.” Following Biblical Aramaic,
ʿlm (or similar forms) occurs in numerous more recent Semitic
languages (Nabatean, Jewish Aramaic, Christian Palestinian Aramaic,
Samaritan, Syriac, Mandaic, Ethiopic, Palmyrene, Egyptian Arabic,
Arabic). Beginning approximately in the 1st century A.D., several of
these languages start using ʿlm in a meaning different from that of
the OT, namely, as “world” or “aeon.”21 Akkadian attests only the
substantively parallel lexeme dārû(m). J. Assmann has shown anew that
the Egyptians did not distinguish clearly between “time” and
“eternity.” They used both the term nḥḥ (the fullness of time,
referring more to what is coming and the change it will bring) and the
term ḏ.t (as consummation, referring more to what endures, abides; the
two words often occur together), referring to a long but finite period
of time as something “unending,” so that here, too, context must
determine whether the translation “time” or “eternity” is more
appropriate. The LXX generally renders ʿôlām (236 times) as aiṓn or
aiṓnios (95 times), less frequently as aeí or archḗ, and 4 times as
chrónos. The previously mentioned occurrence of ʿôlām in various word
and phrase combinations also suggests that the word field associated
with ʿôlām is considerable. Terms include → בוא, bāʾôṯ, → דור dôr, →
יום yôm, hakkōl, → נצח nēṣaḥ, → עד ʿaḏ, → עת ʿêṯ, → קדם qeḏem, and →
II. (Long Ago) Bygone Times. About 60 occurrences of ʿôlām (over 20
with the prep. min) refer to a time long past, or to something
extraordinarily old, albeit with different qualities of remoteness
from the speaker/writer on the one hand, and the observer on the
other. The expression mēʿôlām can mean “from time immemorial” (Ps.
25:6; Jer. 2:20; 31:3; Ezk. 35:5; then Isa. 64:4 (here an isolated
ʿôlām? probably also Josh. 24:2 and the Mesha inscription; cf. Joel
2:2, NRSV “from antiquity”; also Job 22:15: the way of the wicked
“from of old”; and the Aramaic occurrences in Ezr. 4:14, 19).
Mountains and hills are “ancient” (Gen. 49:26; Hab. 3:6), as are gates
(Ps. 24:7, 9; cf. also Jer. 5:15; Ezk. 36:2). Just how old or past the
reference is usually remains open. The point is merely to direct one’s
attention as far back as possible. The most distant time (Ps. 93:2)
can then also refer to an otherwise indeterminate “distant past” (Gen.
6:4) or even to those who died long ago (Lam. 3:6; Ps. 143:3; cf. Ezk.
26:20), or simply an “earlier” time (Josh. 24:2; cf. the expression
“as earlier” in Mic. 7:14; Mal. 3:4). Among the remaining occurrences
(including Dt. 33:15; 1 S. 27:8; Isa. 44:7; 51:9; 63:9, 11; Jer. 5:15;
6:16; 18:18; 28:8; Ezk. 25:15; 26:20; cf. Sir. 16:7; 44
superscription, 2; 48:25; cf. also 42:21 with the article, “one is
from all eternity[?]”; then also 51:8, or “already from an earlier
time”? also Ps. 41:14), Mic. 5:1(2) and Am. 9:11 are noteworthy
inasmuch as they refer this past time to the time of David as the
idealized past. Given Prov. 22:28, one might inquire whether 23:10
should not read → אלמנה ʾalmānâ instead of ʿôlām. In Prov. 8:23
(within the context vv. 22–31) Wisdom remarks in first-person
discourse that she was created by Yahweh even before the creation of
the world, and indeed was herself present at the creation of the
world. This intensifying inclination is also attested in the
combination min- (mē) (hā)ʿôlām (we)ʿaḏ-ʿôlām, which usually in later
texts (cf. combinations with mēʿattâ in Ps. 113:2; 115:18; 121:8;
125:2; 131:3; Isa. 9:6; 59:21; Mic. 4:7) and in the solemn,
liturgically elevated language of prayers and doxologies celebrates
hymnically God’s “eternity” or qualifies such praise as having already
been sung much earlier and, indeed, will be sung much later as well
and ultimately even “for all time”33 (with the article in 1 Ch. 16:36;
Neh. 9:5; Ps. 41:14; 106:48; without the article in 1 Ch. 29:10;
Ps. 90:2; 103:17; Aramaic in Dnl. 2:10; then also in Sir. 39:20,
already as “ages of the world”? further also in the Qumran texts).
When ʿôlām (with min) is negated with reference to the past (Isa.
63:19a; 64:3; Joel 2:2), it expresses the notion “never.”
III. The (Distant) Future. In over 260 instances ʿôlām is used in
reference to the future. In many cases (about 160 times), the
substantive is preceded by le (more common when the reference is to
something static or unchangeable) or (about 80 times) an ʿaḏ (focusing
more dynamically on the temporal progression). The actual “duration”
is often specified as “for always,” “perpetual” (esp. with le) in many
combinations and as an adverbial accusative with largely concrete and
often plural referential words (something also attested by the
parallel expressions; cf. 1 S. 1:11, 22, 28; Ps. 34:1[superscription];
71:15; 89:2; 104:33). Such duration does not, however, necessarily
mean “perpetually, for always,” something attested by 1 S. 2:30f.,
where the time in question must at some point come to an end. This
particular usage, occurring also in texts from the OT environs,38 is
attested in the most varied textual types and periods of OT literature
(Gen. 3:22; 6:3; 13:15; Ex. 3:15; 14:13; 15:18; 19:9; 40:15; Dt. 5:29;
23:4, 7[3, 6]; 28:46: like here and in Gen. 13:15 also in 1 S. 20:42;
2 S. 22:51 par., etc., together with → זרע zeraʿ or → דור dôr or with
the latter in the pl.; then 1 S. 1:22; 20:15; 20:23, 42; Ezr. 9:12; 2
S. 23:5; Ps. 30:13; 49:9; 61:8; 66:7; 73:12; 89:2, 3, 38[1,
2, 37]; 90:2; 106:31; Prov. 27:24, “forever”; Isa. 30:8; 35:10; 55:3;
60:19f.; Jer. 20:7; 23:40; Ezk. 25:15; Jon. 2:7). Job will not live
“forever” (7:16), and the same is asked analogously with regard to the
prophets (Zec. 1:5). The stones in the Jordan will be an “abiding”
memorial (Josh. 4:7). Ps. 77:8(7); 1 K. 8:13; and 2 Ch. 6:2 all speak
of coming times (pl.!). With future reference, negated ʿôlām can mean
both “no longer” (Ex. 14:13) and “never” (1 S. 20:15; Neh. 13:1; Isa.
25:2; Jer. 35:6; Ezk. 26:21; 27:36; 28:19; cf. Dt. 23:4–7[3–6]). The
obvious use of ʿôlām in Ex. 21:6; Dt. 15:17; 1 S. 27:12 (cf. Lev.
25:46; 1 S. 1:22; Job 40:28[41:4]) to mean “as long as one lives”
(e.g., a slave for life) does not necessarily contradict its other
meanings, for even when the king is greeted with “may the king live
forever,” this does not, despite the obvious presence of “courtly
style,” imply the wish that the king be granted eternal life, but
rather that he live “as long as possible” (contrast Job 7:16). When
referring thus to a time enduring long into the future, ʿôlām is quite
naturally and often combined with and intensified or strengthened by
other lexemes. These include dôr (or its pl. or dual; usually with le)
(Gen. 16:7; Ex. 3:15; 31:16; Dt. 32:7; Ps. 33:11; 45:18;
49:12; 61:7f.[6f.]; 77:8f.[7f.]; 79:13; 85:6; 100:5;
102:13; 106:31; 119:89f.; 135:13; 146:10; Prov. 27:24; Eccl. 1:4;
Isa. 34:10, 17; 51:8; Dnl. 3:33[4:3]; 4:31; Sir. 45:26). An
analogous situation already obtained in Ugarit. One additional
intensifying combination is the expression (le) ʿôlām wāʿeḏ (Ex.
15:18; Ps. 9:6; 10:16; 21:5; 37:27 [following LXX]; 45:7, 18[6,
17]; 52:10; 104:5; 119:44; 145:1, 2, 21; Dnl. 12:3; Mic. 4:5; Sir.
40:17), which Jenni calls a “solemn formula of conclusion and
reinforcement” (cf. also the noun → עד ʿaḏ together with ʿôlām in Ps.
111:8; 148:6; Isa. 45:17). The formula mēʿattâ (we) ʿaḏ-ʿôlām is
similarly “solemn” (Ps. 113:2; 115:18; 121:8; 125:2; 131:3; Isa.
9:6; 59:21; Mic. 4:7; cf. Sir. 51:30 [all these are probably later
texts]). Finally, the formula “for his steadfast love endures forever”
([kî] leʿôlām ḥasdô) should be mentioned. Apart from Ps. 136, where it
is a refrain in every verse, it occurs 16 times in the OT (1 Ch.
16:34, 41; 2 Ch. 5:13; 7:3, 6; 20:21; Ezr. 3:11; Ps. 100:5; 106:1;
107:1; 118:1–4, 29; 136 [26 times]; Jer. 33:11; also Sir. 51:12). This
context of reinforcement and intensification also includes the use of
ʿôlām in oaths (Dt. 32:40, divine discourse; cf. Josh. 14:9; Jer.
49:13; Dnl. 12:7; Zeph. 2:9) or in asseveration (2 S. 3:28; 7:26, 29).
The term ʿôlām then also occurs over 120 times in construct
combinations, especially with a future orientation within
theologically significant contexts.
Preuss, H. D. (1999). עוֹלָה and עוֹלָם. G. J. Botterweck, H. Ringgren, & H.-J. Fabry (Eds.), D. W. Stott (Trans.), Theological Dictionary of the Old Testament (Revised Edition, Vol. 10, pp. 531–536). Grand Rapids, MI; Cambridge, U.K.: William B. Eerdmans Publishing Company.